Opinion by Judge RYMER; Section VLB. by Judge N.R. SMITH; Dissent by Judge RYMER.
OPINION
RYMER, Circuit Judge:
These are consolidated appeals from dismissal of toxic tort actions against Whitman Corporation. The district court dismissed a number of plaintiffs on statute of limitations grounds; others for failure to comply with case management orders; some when their expert’s opinion was excluded; and those who brought claims for preconception injury and for intentional infliction of emotional distress on the footing that those claims would not be recognized under California law. The court also taxed costs in an order that is separately appealed. We affirm on all grounds but timeliness of the adult claims, as to which we reverse. Reversal on this ground moots the costs award, which we vacate.
I
In 1945, Joe and R.E. Harrah bought land in Willits, California and began operating a machine shop that changed its name to Remco Manufacturing Co. in 1961 and later to Remco Hydraulics, Inc. The seven-acre site was located at 934 Main Street. The facility did “hard chrome plating,” a process by which a base metal and another substance of a lead alloy are immersed in a tank of chromic acid, and an electric current is passed through the solution, causing hexavalent chromium to adhere to a base metal. Whitman’s predecessor, Stanray Corporation, bought Rem-co on August 21, 1968; Whitman, in turn, sold the company to MC Industries, Inc. on December 16, 1988. The facility closed in 1995 when MC Industries declared bankruptcy.
In 1996, the City of Willits brought a federal action against Remco alleging that the site was contaminated and requesting remediation. People of the State of California and the City of Willits v. Remco Hydraulics, Inc., et al., Case No. CV 96-00283 SI (N.D.Cal.). A consent decree was entered on August 22, 1997, later amended on December 22, 2000. Among other things, the decree established the Willits Environmental Remediation Trust. The Trust has overseen efforts to investigate and remediate the Remco site since 1997. It has taken thousands of air, water and soil samples, and published groundwater monitoring and sampling reports, interim remedial action reports, work plans, and “Fact Sheets” detailing the background of the facility, environmental investigation activities, waste removal programs, findings from testing samples, and remedial actions.
Donna Avila and 100 others who lived in Willits filed suit on August 23, 1999 seeking recovery for medical problems they allege were caused by release of chemicals by Remco. Avila, et al. v. Willits Envtl. Remediation Trust, et al., Case No. C-99-3941 SI (N.D. Cal.).1 A second action, Arlich, et al. v. Willits Envtl. Remediation Trust, et al., was filed on January 16, 2001. These actions were consolidated in district court on February 13, 2002. A third action, Nickerman v. Remco Hydraulics, [832]*832Inc., was brought in state court on December 23, 2004, and removed. Nickerman was consolidated with Avila in June 2008.2 Although the operative consolidated complaint set forth various claims for relief, alleged personal injuries as a result of exposure to contamination from the Remco site are at the heart of the litigation and this appeal.
The litigation spanned ten years in district court. Numerous motions were filed and orders were entered. Not all involved every plaintiff, of whom there were approximately 1000. Hundreds settled. The main appeal focuses on seven rulings that applied to different subgroups, but that resulted in dismissal of all non-settling plaintiffs. The cost appeal is brought by plaintiffs who were involuntarily dismissed but were held jointly and severally liable for deposition costs associated with those who settled.
II
Questionnaires
Case Management Order No. 1, entered April 16, 2001, required answers to a discovery questionnaire on claimed exposure to Remco hazardous substances and alleged injuries from it. Questionnaire Plaintiffs had 270 days to respond. The initial order advised that sanctions, including terminating sanctions, could be sought against anyone who failed to answer in full. At an October 26, 2001 status conference the court extended the deadline to May 31, 2002. When it passed without responses, Whitman moved to dismiss. The court denied the motion on June 17, 2002, and gave Questionnaire Plaintiffs one more extension to July 15, 2002. Its order stated that no further extensions would be granted and warned that, in light of the two extensions already provided, any plaintiff who failed to submit a response by July 15, 2002 would be dismissed.
Meanwhile, at an October 2001 case management conference, the court also ordered responses by January 31, 2002 to a supplemental questionnaire with five questions on the statute of limitations. When that deadline passed without many responses, Case Management Order No. 2 was issued extending the time to March 29, 2002 and warning that anyone who failed to submit a response by that date would be dismissed with prejudice. On April 28, 2002 the court dismissed 218 Supplemental Questionnaire Plaintiffs who had not submitted a timely response, and on July 29, 2002, dismissed the 105 Questionnaire Plaintiffs who had failed to respond by July 15.
Avila argues that this was an abuse of discretion, but we don’t see how. The point was not pressed in district court as to dismissal of those Questionnaire plaintiffs who failed to respond at all, and is waived. In any event, the court granted liberal extensions. Questionnaire Plaintiffs had more than a year to provide responses; Supplemental Questionnaire Plaintiffs had more than five months to answer five questions. They were clearly warned that the court would dismiss any party who failed to file by the extended deadlines. In these circumstances we are not left with a firm conviction of a “clear error of judgment.” In re Exxon Valdez, 102 F.3d 429, 432 (9th Cir.1996) (noting that we review sanctions for abuse of discretion and will not reverse “absent a definite and firm conviction that the district court made a clear error of judgment.”).
[833]*833III
Prima Fade Order
On December 9, 2004 the district court found that, given the litigation had been ongoing for five years, it was reasonable to require those plaintiffs who had never lived in Willits, or who only lived in Willits after December 1988 (when Whitman ceased operations), to make a prima fade showing of exposure and causation. The process was memorialized in Case Management Order No. 4, entered March 1, 2005. It required written statements setting forth “all facts” supporting non-resident and post-1988 resident plaintiffs’ claimed exposure, together with a written statement from an expert describing the condition for which recovery was sought, identifying the chemical to which the plaintiff was exposed, explaining the route of exposure, opining on causation, and setting forth the scientific and medical basis upon which the opinion was based. On October 16, 2006, the court set a December 2006 deadline and on December 6, extended it to February 15, 2007. That deadline was met. Whitman responded on June 22, 2007, moving to strike the report of prima facie plaintiffs’ expert on exposure and causation, Dr. Alan S. Levin. The district court granted this motion on February 6, 2008, finding that the report was wanting under Federal Rule of Evidence 702 and Daubert v. Merrell Dow Pharmaceuticals, Inc., 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993). This left the remaining 41 prima fade plaintiffs unable to show causation, so they were dismissed.3
Prima facie plaintiffs ask us to disapprove the process because, in their view, this type of order bypasses established rules of procedure for discovery and summary judgment. A prima facie order of the sort entered here is sometimes called a “Lone Pine ” order after Lore v. Lone Pine Corp., No. L 33606-85, 1986 WL 637507, 1986 N.J.Super. LEXIS 1626 (N.J.Super.Ct.Law Div. Nov. 18, 1986), in which such an order was issued.4 We have not directly addressed a Lone Pine order before, although we have affirmed dismissals in cases where orders requiring a similar type of prima fade showing were entered. See, e.g., Claar v. Burlington N. R.R. Co., 29 F.3d 499 (9th Cir.1994); Abuan v. Gen. Elec. Co., 3 F.3d 329 (9th Cir.1993). However, the Fifth Circuit Court of Appeals has considered and approved a Lone Pine order in a similar case that also raised issues of exposure and injury from toxic emissions. Acuna v. Brown & Root, Inc., 200 F.3d 335 (5th Cir.2000). We agree with Acuna’s explanation that district judges have broad discretion to manage discovery and to control the course of litigation under Federal Rule of Civil Procedure 16. Id. at 340. In particular, Rule 16(c)(2)(L) authorizes a court to adopt “special procedures for managing potentially difficult or protracted actions that may involve complex issues, multiple parties, difficult legal questions, or unusual proof problems.” No basis appears for us to cordon off one type of order — a prima facie order on exposure and causation in toxic tort litigation — from the universe of case management orders [834]*834that a district court has discretion to impose. We decline to do so.
Neither does the prima facie process skirt accepted procedure. In a case such as this, Daubert obliges a district judge to determine at the outset, pursuant to Federal Rule of Evidence 104(a), whether an expert’s testimony will assist the trier of fact by assessing whether the methodology underlying the testimony is valid and reliable. 509 U.S. at 592-93, 113 S.Ct. 2786. “A trial court not only has broad latitude in determining whether an expert’s testimony is reliable, but also in deciding how to determine the testimony’s reliability.” Elsayed Mukhtar v. Cal. State Univ., Hayward, 299 F.3d 1053, 1063 (9th Cir.2002). Thus, a case management order that focuses on key issues for expert opinion is in aid of the Daubert responsibilities the district judge must discharge.
Prima facie plaintiffs maintain that, even if a Lone Pine order may sometimes be appropriate, it wasn’t here because this case was not complicated, nor did the order conserve time, expense, or judicial resources as it is supposed to do. We do not believe the district court abused its wide discretion in finding otherwise. The case was already protracted, having been pending for five years when the order was entered; it involved many plaintiffs, not all of whom were similarly situated, as well as a defendant that got out of the business in 1988; and it raised difficult issues of proof on exposure and causation. Prima facie plaintiffs had two years in addition to the five that had elapsed to find experts and develop evidence. Also, the order was narrowly tailored to apply to those individuals whose opportunity for direct exposure to contaminants released by Whitman was different from others because they had never lived in Willits or lived there only after Whitman sold the facility. Nor did information called for in the order appreciably duplicate other avenues such as questionnaires. The questionnaires pertained to generalized claims of exposure whereas the prima facie affidavits were to set forth evidentiary facts, and the questionnaire responses themselves indicated that no analysis of health risks or means of exposure had yet been done. Therefore, it was not inappropriate to elicit more specific information through the prima facie process.
Prima facie plaintiffs also suggest that the order impermissibly made discovery unilateral by directing them to make disclosures before Whitman had to do so. However, Federal Rule of Civil Procedure 26(a)(2)(C) gives a district court authority to determine when — and in what sequence — expert testimony must be disclosed. Whitman’s responsive submissions, though scheduled after plaintiffs’, were on a similar timetable. Prima facie plaintiffs made no attempt to depose Whitman’s experts or to obtain a continuance. In these circumstances, and given seven years to develop a case, we cannot say that they lacked a full and fair opportunity to marshal facts in support of their position. See Portsmouth Square, Inc. v. S’holders Protective Comm., 770 F.2d 866, 869 (9th Cir.1985).
Finally, prima facie plaintiffs fault the district court for precluding new fact and expert declarations offered in support of their prima facie showing after the extended deadline of February 15, 2007 had passed. These included a declaration from a percipient witness (Joe Holt), a second declaration from Dr. Levin, and a declaration from a previously undisclosed expert, Dr. Rash Ghosh. They maintain preclusion was improper because the prima facie process threatened no such sanction. The court’s order, however, plainly stated that each of the non-resident and post-1988 resident plaintiffs was to file a written statement on or before the deadline setting [835]*835forth “all facts” in support of their claimed exposure. The court also found that much of Dr. Levin’s second (July 26, 2007) declaration contained statements of opinion and explanations of methodology that could and should have been included in his original prima fade declaration, and that to the extent it purported to respond to Whitman’s papers, the declaration was mostly argumentative. Likewise, the court found that Holt’s declaration contained new evidence — but not newly discovered evidence — that could have been presented by the February deadline. Dr. Ghosh, in turn, opined on a number of matters directly related to the prima facie showing but had not previously been identified as a witness relevant to that showing and, even if considered, his opinions were speculative, based on insufficient facts, and beyond his expertise. Thus, the new declarations were excluded both because they were untimely, and because they were not genuinely rebuttal or otherwise admissible. In this, the court did not abuse its discretion.
IV
Exposure and Causation Discovery
Discovery on exposure and causation for plaintiffs not subject to the prima facie order was on a different track. On November 16, 2007, the district court authorized Whitman to serve the 15 plaintiffs remaining after mediation with discovery on these issues, and ordered them to respond by January 31, 2008. At discovery plaintiffs’ request, the deadline was extended to March 15. The court instructed these plaintiffs to provide expert opinions in response to the discovery if they intended to rely on expert opinions, and warned that if any plaintiff failed to provide a response, “a negative inference and/or. issue preclusion will result.”
After filing responses as directed on March 15, 2008, discovery plaintiffs later sought to add February 2009 and March 27, 2009 declarations by Dr. Levin that, among other things, used a different methodology to assess medical causation; a January 18, 2009 declaration from Dr. Linda Remy, a statistician and epidemiologist, indicating that Willits had more hospital admissions for serious illnesses than the rest of Mendocino County; and a May 22, 2008 declaration by Dr. William R. Sawyer, which discussed blood tests and sought to bolster Dr. Levin’s opinion about dioxin exposure and injury. The court struck Dr. Levin’s February 2009 report and March 27, 2009 declarations because both should have been produced on time. It also found Dr. Sawyer’s May 22, 2008 declaration untimely, noting that no explanation was provided for why the opinions expressed in it were not expressed in his March 17, 2008 report and that, regardless, the later opinions (on dioxin creation and blood test results) could and should have been furnished by the deadline because they were not based on any new facts.
Discovery plaintiffs posit infirmity in the discovery process because it imposed unilateral expert discovery on plaintiffs, but this argument fails for the reasons we have already explained. More specifically to the discovery process, they contend they should have had more time, and should have been allowed to supplement their responses after the March 15, 2008 deadline. By this time, discovery plaintiffs had had eight years to develop their case. Although they emphasize the need to revamp once the court struck Dr. Levin’s prima facie report on February 6, 2008, the point is not compelling because discovery plaintiffs did not ask for an extension. They further contend that evidence presented after March 15, 2008 was timely because the overall expert cutoff date was [836]*836January 18, 2009, but this, too, is unpersuasive. As the district court explained, expert evidence on exposure and causation was due by the March 15, 2008 deadline whereas the parties had until January 18, 2009 to designate experts on other subjects. The district court is the best judge of its own orders, see United States v. Angelini, 607 F.2d 1305, 1308 (9th Cir.1979), and its interpretation is, in any event, a reasonable one.
Finally, discovery plaintiffs suggest that the declarations as well as the late submission of blood test results5 were a supplement to their prior response which, they assert, they had a right to do under Rule 26(e). However, they do not claim that these submissions were meant to correct or complete a prior disclosure, see Fed.R.Civ.P. 26(e), and the district court found to the contrary that the new reports were not merely refinements of previous reports, but were entirely new. With the defense and the court focused on the exposure and causation evidence produced on March 15, 2008, and with trial approaching, it cannot be gainsaid that changing course at such a late stage would be harmful. We cannot fault the district court’s exercise of discretion in these circumstances.
V
Rule 702 /Daubert Rulings
In California, “in a personal injury action causation must be proven within a reasonable medical probability based upon competent expert testimony.” Jones v. Ortho Pharmaceutical Corp., 163 Cal.App.3d 396, 209 Cal.Rptr. 456, 460 (1985). Plaintiffs must also establish that the substance at issue was capable of causing the injury alleged (general causation), and that the substance caused, or was a substantial factor in causing, the specific plaintiffs injury (specific causation). See, e.g., In re Hanford Nuclear Reservation Litig., 292 F.3d 1124, 1133 (9th Cir.2002); Jones, 209 Cal.Rptr. at 462. There can be no causation if there has been no exposure. Ferris v. Gatke Corp., 107 Cal.App.4th 1211, 132 Cal.Rptr.2d 819, 826 n. 4 (2003).
Federal Rule of Evidence 702 provides that expert testimony is admissible “[i]f scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue”; the witness is qualified “by knowledge, skill, experience, training, or education”; and “(1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.” District judges are the gatekeeper for expert testimony to assure that it is both relevant and reliable. Daubert, 509 U.S. at 592-93, 113 S.Ct. 2786.
Avila offered Dr. Levin as her medical and toxicology expert. He submitted a declaration on February 15, 2007 to comply with Case Management Order Nos. 4 and 6 (the prima facie orders), and a March 15, 2008 declaration to respond to the January 23, 2008 discovery order applicable to non -prima facie plaintiffs. The district court struck Levin’s February 2007 report in a February 6, 2008 order; this ultimately led to dismissal of 41 prima facie plaintiffs for they could not show causation. The court also struck Levin’s March 15, 2008 declaration and granted summary judgment against the ten remaining non -prima facie plaintiffs in a comprehensive order entered June 18, 2009. Together, these Levin plaintiffs [837]*837contend that the court abused its discretion under Daubert.
Levin’s prima facie submissions opined that evidence that halogenated hydrocarbon industrial waste products were burned at the Remco sites over the years “makes it scientifically impossible to conceive that dioxins and PCBs were not emitted. Stated more succinctly, there is no plausible scientific excuse to ignore the unquestionable presence of dioxins on the Remco contamination sites.” He also stated that a groundwater plume containing trichlorethylene (TCE) and other volatile organic chemicals (VOCs) had spread throughout Willits.6 Levin did not conduct any soil, air, or water tests at or around the Remco site, nor did he cite any studies showing dioxin in or around the site. However, Levin had the blood of four plaintiffs, one of whom was part of the prima facie process, tested for the presence of dioxin congeners. From the blood results he extrapolated that all prima facie plaintiffs were exposed to “excessively high levels” of dioxins, and to an unknown amount of a combination of chemicals (PCBs, dioxins, TCE) and VOCs that caused their injuries.
The district court concluded that Levin’s prima facie opinions did not meet the requirements of Federal Rule of Evidence 702. Among other things, the court found that Levin lacked scientific expertise to state that burning activities could or did result in the creation of toxicologically significant amounts of dioxins; there was no actual evidence that dioxins have been present at the site, and his opinion that burning industrial waste created dioxins was based on speculation; Levin’s blood analysis was flawed because 80% of the dioxin congeners were non-detects, yet Levin assumed that non-detected congeners were actually present at the limit of detection; and he failed to eliminate other sources of potential dioxin exposure, such as a medical waste incinerator, traffic, cigarette smoke, and burn barrels. Additionally, the court found there was no scientific support in the documents mentioned in Levin’s declaration for his opinion that a combination of chemicals caused the injuries. Accordingly, it concluded that Levin’s ultimate opinions were inadmissible.
In his March 15, 2008 declaration Levin stated that review of documents available at the Willits Trust depository, the January 15, 1998 Versar Preliminary Removal Site Evaluation Report, discovery responses and depositions relating to burning activity at Remco, and blood tests, demonstrates that toxic chemicals were in use at the Remco facilities and leaked out, contaminating the ground, water, and air of inhabitants in the area. He opined that the mix of chemicals emanating from the Remco site contained carcinogens, and immunotoxic and neurotoxic materials. He also noted there are other “possible” sources of exposure, but “no other proven probable sites that can be factored into this causation analysis.” In addition, Levin expressed the opinion that, based on a clinical differential diagnosis,7 the Remco mix is a biologically plausible cause compounded by an appropriate temporal asso[838]*838ciation between the exposure and development of the disease, without confounding factors; therefore, some single chemical or unique mix of chemicals emanating from the Remco site were substantial contributors to each of the remaining non-prima facie plaintiffs injuries.
Ruling on Whitman’s motion to strike Levin’s March 15, 2008 report and for summary judgment on exposure and causation, the district court noted that the 2008 declaration presented no new evidence or science in support of Levin’s dioxin theory. To the extent plaintiffs claim to have new evidence of burning, the court reiterated its earlier finding that Levin lacks scientific expertise to state that burning activities could or did result in the creation of toxicologically significant amounts of dioxins. The court also observed that if there had been exposure, it would be reflected in the blood test results. For reasons explained in its February 2008 order, the court found that Levin’s opinions based on blood test results were still fundamentally flawed even though Levin reduced by half the level he used for non-detects. The court also rejected Levin’s opinion on exposure to VOCs through soil gas volatization because his 2008 theory was the same as his unsupported 2007 theory. The court again found no facts showing that the groundwater plume extended beyond the plume identified and documented by the Willits Trust. It observed that the sources on which Levin relied to show a broader plume were preliminary, and do not support Levin’s claim that the areas of the plume where levels of TCE or VOCs in the groundwater exceed the EPA’s maximum concentration level for drinking water covered a “large geographic area.” With respect to Levin’s March 2008 opinion that hexavalent chromium caused certain of the Levin plaintiffs’ injuries, the court found other evidence, in particular, the 2004 Final Public Health Assessment for Abex/Remco Hydraulics Facility, showed they lived' in Willits at a time when the maximum estimated concentration was five times less than the threshold for causation. Finally, the court found that Levin did not distinguish between emissions before and after December 16, 1988, when Whitman ceased operations, and that the failure was fatal absent a showing that Whitman was responsible for post-1988 exposures. Thus, the court held, Levin’s opinions on exposure are speculative and unreliable.
Regardless, even if Levin could establish exposure, the court concluded that his 2008 declaration failed to show either general or specific causation. In sum, the court found no relationship between the asserted effects of chemicals and plaintiffs’ injuries.8 Nor was Levin’s report reliable enough to support specific causation, because he simply assumed that causation existed without going through the steps in the differential diagnosis process that his report said he would. Missing evidence of exposure, and general or specific causation, Levin plaintiffs could not make out a case. Accordingly, the court granted summary judg[839]*839ment to Whitman against the remaining non -prima facie plaintiffs.
All Levin plaintiffs maintain that Dr. Levin met the minimal education or experiential qualifications to give an opinion as to whether burning of halogenated hydrocarbons (solvents) creates dioxins. However, we do not believe the district court’s assessment was clearly in error. Levin described himself as “a Physician/ScientisVAttorney,” and indicated that his areas of scientific expertise include cancer immunology and biology, basic and clinical immunology, and medical toxicology. He has degrees in chemistry, but nothing in the record indicates that Levin had any special training or knowledge regarding metal working industries such that he could reliably opine that activities at Remco “must” have created dioxins. Levin plaintiffs rely on Ralston v. Smith & Nephew Richards, Inc., 275 F.3d 965, 969—70 (10th Cir.2001), to argue that lack of specialization goes to weight, not admissibility, of an expert opinion; but Ralston does not embrace this proposition without qualification. Rather, as it states, lack of specialization may go to weight only “as long as an expert stays within the reasonable confines of his subject area.” Id. We see no abuse of discretion in the court’s determination that an opinion on the content of waste in the chrome-plating industry and what flows from it are outside the areas of Levin’s expertise. For this reason alone, his opinions were properly excluded for lack of qualification. This being so, we have no need to reach whether Levin’s conclusions are reliable.
Nevertheless, the district court’s analysis on reliability is well founded. In short: Even considering testimony by several witnesses that burning occurred on some occasions at Remco, there is none upon which Levin relied that any particular type of solvent was used. So far as blood work is concerned, the lab to which the samples were sent tested for the presence of 17 dioxin congeners. Overall, 80% of the dioxin congeners were not detected, that is to say, Avila and another sampled plaintiff had non-detects for 14 of 17, a third test plaintiff had non-detects for 12 of 17, and Nickerson, 15 of 17. But Levin assumed that the dioxins were present at the detection limit, then converted the results to an assumed “lipid-adjusted” basis without determining from the lab what that basis actually was, and compared those results with results from two published studies to conclude that the blood test plaintiffs had high levels of dioxin in their blood and that the signature congener profile showed exposure from a single source. As the district court found, Dr. Levin’s method of creating a congener profile had no scientific support. See Daubert v. Merrell Dow Pharmaceuticals, Inc., 43 F.3d 1311, 1319 (9th Cir.1995) (relying on an expert’s bare assertion that the methodology employed comports with standard scientific procedures is insufficient under Daubert). Likewise, Levin’s treatment of other potential sources of dioxin falls short because it does not actually consider confounding factors. His February 2007 report failed to consider and rule out other sources at all; his 2008 report did assert the absence of any other proven “probable” sites, but offered no explanation for why or how he eliminated other sources of dioxin. While an expert opinion does not necessarily have to eliminate all causes of injury to be admissible, Levin’s entire theory turned on Remco’s being the only possible source of the dioxin levels that he assumed were present in the blood of the four test plaintiffs.
To the extent Levin’s opinion is premised on exposure to volatile organic [840]*840compounds,9 there is no evidence that VOCs are in the .blood of any plaintiff; there is no evidence that VOCs were present in any air, soil, or water samples, and Levin did no testing of his own; he did not explain how VOCs traveled from the groundwater through the soil; and he used preliminary estimates of the size and scope of the underwater plume. Similarly, to the extent he opined in his 2008 report (though not in his 2007 report) that exposure to hexavalent chromium contributed to injuries of the Levin plaintiffs, it was unreliable because it relied on the higher maximum exposure level reported in the 2003 draft Public Health Assessment for Abex/Remco Hydraulics Facility instead of the 2004 Final Public Health Assessment for Abex/Remco Hydraulics Facility. The dose estimates by Dr. Sawyer, another of Levin plaintiffs’ experts, were similarly unreliable because they, too, were calculated on the basis of the 2003 draft.
Levin plaintiffs argue that the district court rejected differential diagnosis as a methodology, but that’s not what the court did. It explicitly declined to decide whether differential diagnosis in general was adequate, because Levin did not follow his own four-step procedure. For example, he listed injuries without finding whether those injuries had in fact occurred, whereas his first step was to “identify the valid presence of disease.” And he offered no specifics to show a plausible link between disease and exposure, or between exposure and onset of disease. . Nor did Levin’s report “consider” confounding factors; it just dismissed them. Although Levin cited peer-reviewed articles, there is no indication that any of them connected either the “mix” of chemicals, or any single chemical, with the Levin plaintiffs’ injuries. Dr. Remy’s report doesn’t save the day, for she characterized it as “preliminary” and it also fails to tie any increased incidence of medical disorders in Willits with any of the hazardous substances at issue. Finally, Levin treated the period from the early 1950s through 1995 as one continuous block despite the fact that Whitman stopped operating the facility in December 1988. While Whitman would be responsible for chemicals released at Remco prior to then, it would not be responsible for releases from 1988 through 1995. Thus, Dr. Levin needed to consider and rule out exposure to Remco chemicals after 1988, but did not do so.
VI
Statute of Limitations
The court granted summary judgment on statute of limitations grounds as to 15 minor and incompetent plaintiffs to the extent their claims were for birth defects; 160 adults (limitations plaintiffs), because they failed to show the discovery rule applied to claims as to which the statute of limitations had otherwise expired; and Brenda McCann on her action for the wrongful death of her father, because a judicial admission disclosed that she knew or should have known about the cause of his death more than a year before filing.
a.
Minor plaintiffs argue that they are entitled to tolling pursuant to California Code of Civil Procedure § 352(a). However, this argument was not made in district court and is waived.10 In any event, it is unresponsive to Whitman’s motion for [841]*841summary judgment, which sought dismissal of birth defect claims. Such claims involve a different limitations period and are governed by a different section of the Code of Civil Procedure, § 340.4. This ground was uncontested.
N.R. SMITH, Circuit Judge:
B
On March 13, 2003, the district court granted summary judgment against all Plaintiffs filing suit after August 24, 2000. We hold that this was error and vacate the grant of summary judgment. The district court correctly held that this issue may be decided on summary judgment. Summary judgment is appropriate when the material facts are undisputed and the moving party is entitled to judgment as a matter of law. Fed.R.Civ.P. 56(a). Here, the events and publicity in and around Willits are undisputed. It is also undisputed for purposes of summary judgment that Plaintiffs knew of their injuries more than a year before filing suit. The only issue is a matter of law — considering the evidence, should Plaintiffs have known of the cause of their injuries more than a year prior to filing suit.
Under CERCLA’s provisions, a limitations period does not begin until the date “that [a plaintiff] knows or reasonably should have known” of both the existence and the cause of the injury. O’Connor v. Boeing N. Am., Inc., 311 F.3d 1139, 1147-48 (9th Cir.2002). California’s statute of limitations for personal injury at the time of suit was “within one year.” Cal.Civ.Proc.Code § 340(3) (2001). Therefore, summary judgment was proper if the only “reasonable inference that can be drawn is that Plaintiffs knew or should have known more than one year before filing their claims that the [Remco] contamination was the cause of their diseases.” O’Connor, 311 F.3d at 1149-50. In other words, Plaintiffs had one year from the date that they “should have known” of their cause of action to file their lawsuits.
After detailing events and publicity in Willits and the surrounding area between 1979 and August 2000, the court found that “ ‘the only reasonable inference’ to be drawn is that all plaintiffs knew or should have known by August 24, 2000, one year from the date on which Ms. Avila filed her suit, that their alleged injuries were caused by the Remco Facility. Therefore, the Court GRANTS defendant’s motion for summary adjudication as to the 359 plaintiffs who filed after August 24, 2000[.]” This is erroneous as a matter of law. The statute of limitations begins to run on the date by which the Plaintiffs knew or should have known of their injuries. Therefore, summary judgment would only be appropriate for those Plaintiffs filing one year from August 24, 2000— the date the district court found they “should have known” that their alleged injuries were caused by the Remco Facility.
Both parties apparently assumed that the district court meant that Plaintiffs should have known of their injuries a year before August 24, 2000, i.e., as of the filing of Avila’s suit on August 23, 1999. While such an interpretation would be a valid application of the law, it is contradicted by the facts articulated in the district court’s order. To begin with, the district court states “after the Avila suit was filed on August 23, 1999, there was substantial resulting publicity about the health risks posed by the contamination from the Rem-co site.” In addition, the district court specifically cited Avila’s health survey (conducted in September 1999), and Erin Brockovich and Ed Masry speaking in Willits in August 4, 2000. If Plaintiffs should have known of their injury and its cause as [842]*842of August 1999, such publicity would be irrelevant.
Furthermore, holding August 24, 1999 as the date by which Plaintiffs should have known of the date of their injuries is not supported by the record before us. Information on Remco prior to August 1999 focused primarily on legal issues and the contamination. Very little information indicated that Remco could be the cause of Plaintiffs’ alleged injuries. Publicity concerning the Remco contamination and that it was likely the source of that injury must have been so pervasive as of that date that the only reasonable inference is that Plaintiffs should have known the cause of their injury. The record before us simply does not support such an inference. The district court stated that
it is plausible that many residents in the City of Willits remained unaware for a long period of time of the health risks caused by the contamination by the Remco plant, particularly because, as plaintiffs note, many local newspaper articles specifically rejected any allegations of health risks in other reports.
As the court noted, it was only after Avila’s suit that there was considerable publicity about the health risks, although the publicity was still equivocal. For example, after Avila’s filing the local paper reported that Avila and others were suing for “medical monitoring” because the children at Blosser Elementary were “always getting sick” and that Avila and others wanted the school district to test the water at Blosser. Followup articles claimed chromium contamination was limited to a few areas on the Remco site and confirmed that the water at Blosser Elementary was safe.
The first discussions about health problems actually being caused by Remco occurred just prior to the October 19, 1999 town meeting. For the first time there was an extensive discussion of the contamination, methods of exposure, and the apparently large numbers of people in the town with cancer. Reports of the October 19, 1999 town meeting, however, noted that it was clear that no one knew the extent of the contamination. In November and December 1999, the newspaper noted that health surveys were planned, but did not discuss the types of injuries associated with chromium, except cancer.
A January 15, 2000 meeting focused on health issues — specifically, the incidences of cancer, hysterectomies, birth defects, and nosebleeds. Avila presented findings from her health survey, and an attorney spoke about the hazards of chromium contamination. However, newspaper reports pointed out that many allegations were made, but not only were they scientifically lacking, some had been scientifically contradicted.
After the January meeting, articles and meetings were more numerous and more focused on health issues, and the public debate became more widespread. The news and health studies continued to be equivocal and contradictory. However, the news stories show that health concerns were at the forefront of the public debate. For example, in January and February various cancer studies found no statistically significant increase in cancer rates in Willits. The draft Preliminary Endangerment Assessment (PEA) claimed that the health threat from Remco contamination was insignificant, but a technical review noted that the study showed “that there are potential long-term effects off-site, particularly if residents are exposed to contaminated ground water.” The PEA gave specific information about various paths of exposure and the attendant risks, although not all were fully explored. It also had statistics on cancer and general health risks based on the residence of the [843]*843individual. Residents and citizen groups called for epidemiological studies to determine the health effects of the contamination.
Public events also became more numerous. Meetings with attorneys for “all persons who have an interest” in the Remco contamination, announced by large paid advertisements, were held on April 15, 2000 and June 10, 2000. On May 18, 2000, the town held a meeting regarding the PEA, and on July 14, 2000, a Citizens’ Advisory Committee meeting addressed the health hazards of even low levels of toxins. In addition, the state Department of Health Services spent three days in Willits talking with concerned citizens. In July, it began a health survey of past and present health risks associated with the contamination, which was slated to take between three months and a year to complete.
Coverage of the possible health problems in Willits also became more widespread during this time.11 The Santa Rosa Press Democrat began to carry stories of the contamination, and the Ukiah Daily Journal increased its coverage. The San Francisco Chronicle also covered the contamination, specifically the health effects, in a large multi-page story “A Town Diseased and Torn.” The hazards of chromium exposure gained nationwide attention beginning in March 2000, with the release of the movie Erin Brockovich. Indeed, one article compared the situation in Willits with the movie. It noted the health problems Brockovich discovered in Hinckley, California (stomach ulcers, miscarriages, hysterectomies, kidney disease, etc.) and states that “chromium 6 is a deadly poison.” As the district court noted, Brockovich became involved in the Willits contamination soon after the movie was released, even holding an information session in Willits on August 4, 2000.
The ever-increasing interest in health issues associated with the Remco contamination (evidenced in the excerpts of record on appeal) demonstrates that by August 24, 2000, people with a reasonable degree of awareness should have known that contamination from Remco was a serious health concern and made further inquiry. Therefore, Plaintiffs should have known that Remco was the likely cause of their injuries no later than August 24, 2000. However, not having the full record before us, we decline to determine a precise date by which they should have known. We therefore (1) hold that as a matter of law, the events leading up to August 24, 1999 were insufficient for the court to conclude “the only reasonable inference” was that Plaintiffs should know of their cause of action as of that date, and (2) vacate the court’s grant of summary judgment dismissing all claims filed after August 24, 2000.
C
John Harrington died on October 10, 1992, of kidney cancer. His daughter, Brenda McCann, filed a wrongful death action on January 21, 2002. Although McCann asserts that she assumed until the spring of 2001 that her father’s cancer and death were attributable to smoking, the discovery rule does not save the wrongful death claim because her individual complaint, filed on August 23, 1999, avers that hexavalent chromium was a known carcinogen. Also, she indicated that it was questions about her family’s health history and cancer that led her to believe that her father’s cancer had been caused by exposure to Remco chemicals; but similar questions were in Avila’s health survey, in [844]*844which McCann participated. As such, she knew or should have known that cancer could be related to Remco more than a year before she filed. Accordingly, the court did not err in granting summary judgment on her claim.
VII
Preconception Claims
Thirty-eight plaintiffs seek compensation for injuries allegedly caused by the chemical exposure of a parent, grandparent, or great-grandparent, before they were conceived. The district court concluded that Whitman owed no duty of care to these subsequently conceived offspring. The court acknowledged that a preconception tort cause of action was recognized in Turpin v. Sortini, 31 Cal.3d 220, 182 Cal.Rptr. 337, 643 P.2d 954 (1982), but correctly, in our view, believed that California would not extend the tort to these particular claims. In Turpin, a medical care provider negligently failed to discover that Turpin’s oldest daughter suffered from an hereditary ailment; as a result, Turpin had another child afflicted with the same condition. In this context, the court found that a preconception claim was cognizable. Since then, however, Turpin has been limited to its facts. See, e.g., Foy v. Greenblott, 141 Cal.App.3d 1, 190 Cal.Rptr. 84, 93-94 (1983) (construing Turpin to permit preconception claims only where defendants failed to diagnose and warn parents of the probability an infant will be born with a hereditary disability). Hegyes v. Unjian Enter., Inc., 234 Cal.App.3d 1103, 286 Cal.Rptr. 85 (1991), upon which preconception plaintiffs rely, is not to the contrary. It notes that ho duty has been found, nor has any liability been imposed, in a preconception negligence case where the defendant was not a medical professional or-.product liability manufacturer. Id. at 89. This remains so to this day. As Whitman provided no goods or services related to the reproductive process, we cannot say that it owed a duty to subsequently conceived offspring.12
VIII
Claims for Intentional Infliction of Emotional Distress
Avila also brought a claim for the intentional infliction of emotional distress, on which the district court granted summary judgment for lack of a showing that Whitman actually knew of these particular plaintiffs. She argues that this is wrong because Whitman spilt chemicals in the middle of town and knew what it was doing to the residents. However, the district court’s conclusion follows from Potter v. Firestone Tire & Rubber Co., 6 Cal.4th 965, 25 Cal.Rptr.2d 550, 863 P.2d 795 (1993), the controlling California authority. Potter made clear a plaintiff must establish that the defendant engaged in egregious conduct directed toward the plaintiff. In so holding, the California Supreme Court rejected essentially the same argument Avila advances here, that the defendant “had to have realized that its misconduct was almost certain to cause severe emotional distress to any person who might foreseeably consume the water and subsequently discover the facts.” Id. 25 Cal.Rptr.2d 550, 863 P.2d at 820. This did not satisfy the knowledge requirement in Potter, and doesn’t here.
IX
Costs
Non-settling plaintiffs separately appeal two aspects of the district court’s award of [845]*845costs: taxing costs associated with plaintiffs who settled their claims, and imposing joint and several liability. This issue is mooted by our reversal on the statute of limitations.
X
Conclusion
We conclude that the district court had discretion to require responses to questionnaires, responses to discovery, and a prima facie showing on exposure and causation as to selected plaintiffs, and to enforce its orders by dismissal when responses were not forthcoming by extended deadlines. Likewise, the court had discretion to find that Avila’s expert on exposure and causation lacked expertise to opine on whether burning solvents at Remco could or did result in the creation of toxicologically significant amounts of dioxins. In any event, the district judge had latitude to exercise her gatekeeping responsibilities under Daubert by excluding opinions that were not supported or reliable.
The court did, however, err in granting summary judgment on statute of limitations grounds as to the adult plaintiffs. As California law does not extend to the type of claim for preconception injury brought by certain plaintiffs, or to a claim for intentional infliction of emotional distress against a defendant that didn’t target the particular plaintiffs, the court properly dismissed both of these claims.
Finally, the costs award is moot, as we have reversed on the statute of limitations. We therefore vacate it.
AFFIRMED IN PART; REVERSED IN PART; VACATED IN PART.