RYMER, Circuit Judge:
Raymond E. Andrews appeals the district court’s entry of summary judgment for the Secretary of Health and Human Services, affirming the Secretary’s denial of Supplemental Security Income (SSI) benefits under Title XVI of the Social Security Act, 42 U.S.C. § 1381 et seq. Andrews argues that the Administrative Law Judge improperly rejected the conclusions of the examining psychologist without giving clear and convincing reasons for doing so, and that the determination of nondisability was not based on substantial evidence. 42 U.S.C. § 405(g). He further asserts that, even if the ALJ were entitled to discount the opinion of the psychologist who examined him, the ALJ’s decision must nonetheless be reversed because the hypothetical that formed the basis for the vocational expert’s opinion did not accurately reflect the limitations found by the medical advisor. We have jurisdiction pursuant to 28 U.S.C. § 1291. We hold that the Secretary was entitled to adopt the opinion of the nonexamining medical advisor, who was present at the hearing and testified, and to discount the opinion of the examining physician, because the ALJ gave specific and legitimate reasons for doing so that were based on substantial evidence in the record in addition to the nonexamining psychologist’s opinion. However, because we agree that the hypothetical given to the vocational expert was inadequate, we reverse and remand for further proceedings.
I
Andrews was born in 1964. He dropped out of high school but secured an equivalency certificate. Andrews has worked as an electrical machine winder, a pizza cook, a craftsman for a stained glass manufacturer, a carpenter/laborer, and a machine off-bearer. Andrews most recently worked as a cook in a restaurant, but by his own account quit after only half a day because of the stress involved. Andrews was homeless at the time of the hearing before the ALJ, living at a “mission” and doing cleaning and salvage work in exchange for his daily room and board.
Andrews suffers from psoriasis which flares up periodically. Consequently, the ALJ found that jobs involving exposure to certain caustic substances and detergents are unsuitable. Andrews’s appeal, therefore, challenges only the ALJ’s view of his psychological problems: drug addiction, paranoia (schizotypal personality), adjustment disorder and depression, all allegedly resulting in an inability to function in social situations. Andrews was found disabled by the state of Oregon for the purpose of qualifying for state public assistance, and has been rejected for vocational rehabilitation because of his drug use.
In December 1987, Andrews was diagnosed as having a dysthymic mental disorder and a mixed substance abuse disorder. On September 30, 1990, Andrews presented in a hospital emergency room and was diagnosed with alcoholism, a mixed personality disorder, and an adjustment disorder with mixed features. William McConochie, Ph.D., a licensed psychologist, examined Andrews on four occasions in connection with Andrews’s application for public assistance benefits- provided by the state of Oregon. After the first examination, on November 6, 1990, McCono-chie concluded that Andrews did not “appear to be actively psychotic” and that his thought processes were “generally intact.” He noted
that Andrews was “somewhat paranoid.” The diagnosis included major depression, recurrent, severe; and a schizotypal personality disorder. McConochie rated Andrews’s prognosis as “poor.” McConochie credited Andrews’s claim that he had his drug problem under control. McConochie did not report conducting any standardized psychological or intelligence tests.
McConochie examined Andrews again on April 3, 1991. Despite noting a general improvement in his mental condition, McCono-chie found him to be “unemployable for the next six months on the open job market.” McConochie found that Andrews would benefit from work in a “semi-sheltered” setting. He examined Andrews for a third time on October 8, 1991. He noted Andrews’s paranoia, anxiety and drug dependence and stated that Andrews was moderately depressed. The diagnosis was cannabis dependence, post-traumatic stress disorder based on an abusive childhood, and schizotypal personality disorder (relating to paranoia and anxiety). The prognosis was rated as “poor” and “chronic.” McConochie noted that Andrews had “many excuses” for why he could not seek counseling or attend school.
On December 16, 1991, McConochie filled out a “Medical Assessment of Ability to do Work-Related Activities (Mental).” On a four-point scale that ranged from “Unlimited/Very Good” to “Poor/None,” McConochie rated Andrews’s various employment related attributes as “Poor/None” in 12 of 15 categories. McConochie opined that Andrews was “severely handicapped by chronic drug abuse, depression and paranoid ideation.” A December 31, 1991, evaluation yielded comments substantially similar to those made in the October evaluation.
Andrews testified that he felt anxious and trapped in employment situations. He asserted that he had sought help with his mental health, but indicated a lack of resources to pay for a program and that “there aren’t any free programs for males.” Andrews testified that he had been “clean and sober” prior to October, 1991, but by that time began having anxiety attacks and “used marijuana as needed to control my anger and aggressive feelings.” He said that, at the time of the hearing, he was using neither alcohol nor drugs; that he had no desire to smoke marijuana; and that he wanted to drink, but did not. Andrews testified that he had previously smoked about a gram of marijuana a day for a two week period in 1991, although McConochie’s reports from this period indicate a longer period of use. Andrews testified that he had stopped drinking around December, 1990; however, he told McCono-chie in November, 1990, that he had stopped drinking 18 months earlier.
According to Andrews’s testimony, he briefly shared an apartment with a woman that he was “trying to have a relationship with.” Another brief relationship ended after the woman was diagnosed with “multiple personality disorder.” Andrews stated that he regularly attends local Alcoholics Anonymous meetings and stays until closing at 10:00 PM, when he goes to the mission. He stated that he has no problem sleeping. Andrews testified that he considers the police a potential enemy and that he feels threatened by the “movement” against the right to bear arms, but stated that he didn’t think “people are following me around.” Andrews stated that despite the hostile environment he experienced on the streets, “I’m not out there stealing and I’m not out there killing people for money and I still have the desire to succeed if I’m given a chance.” Andrews stated that he had attempted suicide in the past but now doesn’t “think about it in the sense where I’m going to carry it out today.” He testified that he went to Sacred Heart Hospital on two occasions after anxiety attacks; the visits were unsuccessful because Andrews perceived the counselor as hostile. Andrews noted that he had learned “manipulation” on the streets, and that when he would go before a court he would “play stupid” to “get leniency.”
Andrews’s medical files were reviewed by five mental^Jiealth professionals, including two psychiatrists, for purposes of determining disability. None rated him as more than moderately impaired in any category on the mental residual functional capacity (MRFC) form. One, Janice Green, Ph.D., a psychologist with special experience in the treatment of substance abuse, testified at the January
21, 1992, disability hearing before the ALJ. Green did not examine Andrews in a clinical setting, but she had reviewed Andrews’s medical records and heard him testify, Based on these observations, she expressed the opinion that Andrews’s primary disability was polysubstance abuse; that Andrews was able to control his drug use and that, in light of his activities, the abuse was in remission; that Andrews’s medical record was based on self-reporting; that there probably was a personality disorder, but the record did not support a diagnosis of schizophrenia; and that his work limitations would consist of slight restrictions in daily activities, minor difficulty in social function, regular occurrences of deficiency of concentration, persistence and pace, and one or two episodes of deterioration. Her assessment of Andrews’s mental residual functional capacity concurred with that made on July 9, 1991, by K. Bates Smith, Ph.D., which was based on a review of documentary evidence, including McCono-chie’s evaluations to that date. According to this evaluation, Andrews was no more than moderately limited in any of the 20 standard categories. Green also opined that the results of psychological testing would be clouded by Andrews’s substance abuse and thus unreliable.
The ALJ found that Andrews was not an uncontrollable alcohol or drug addict and that he was “a very skillful, manipulative person who has learned to exploit persons and situations to his best advantage.” Because MeConochie’s opinion as to disability was made “without consideration of the claimant’s own admission that he is a manipulator and therefore not a generally credible person,” and because the substance abuse rendered the psychological diagnosis unreliable, the ALJ rejected MeConochie’s conclusion that Andrews was disabled and his assessment of Andrews’s residual functional capacity. Finding that Andrews was not a “medically determined alcoholic or drug addict,” as defined in 20 C.F.R. § 416.935, since he had not lost the voluntary ability to control his substance use, the ALJ concluded that Andrews was not disabled.
The ALJ also found that, because of his psoriasis, Andrews had established his inability to perform his past work. Having rejected Andrews’s mental impairment claims, the ALJ relied on the vocational expert’s answer to a hypothetical involving only Andrews’s psoriasis and a need to have “minimal interaction” with other people. Based on this hypothetical, the vocational expert concluded that the claimant could still perform 150,000 to 175,000 jobs in the national economy. Accordingly, the ALJ found that Andrews was not disabled and denied his application for benefits. The Appeals Council denied review, and the decision of the ALJ became the final decision of the Secretary.
Andrews filed an action for review of the Secretary’s decision in district court pursuant to 42 U.S.C. § 1383(c)(3). By consent of the parties, a magistrate judge heard the case and entered final judgment. 28 U.S.C. § 636(b)(1)(C). The magistrate judge granted summary judgment in favor of the Secre: tary, affirming her decision. Andrews now appeals.
II
“The findings of the Secretary as to any fact, if supported by substantial evidence, shall be conclusive_” 42 U.S.C. § 405(g). The Secretary’s decision to deny benefits will be disturbed only if it is not supported by substantial evidence or is based on legal error.
Magallanes v. Bowen,
881 F.2d 747, 750 (9th Cir.1989).
Substantial evidence means more than a mere scintilla but less than a preponderance; it is such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.
Id.
To determine whether substantial evidence supports the AL J’s decision, We review the administrative record as a whole, weighing both the evidence that supports and that which detracts from the ALJ’s conclusion.
Id.
The ALJ is responsible for determining credibility, resolving Conflicts in medical testimony, and for resolving ambiguities. '
Id.
We must uphold the ALJ’s decision where
the evidence is susceptible to more than one rational interpretation.
Id.
The Secretary has established a five-step sequential evaluation process for determining whether a person is disabled.
Bowen v. Yuckert,
482 Ü.S. 137, 140, 107 S.Ct. 2287, 2290-91, 96 L.Ed.2d 119 (1987); 20 C.F.R. § 416.920. The claimant bears the burden of proving entitlement to disability benefits.
Gallant v. Heckler,
753 F.2d 1450, 1452 (9th Cir.1984). If, as in this case, the claimant shows that he is unable to perform past relevant work, the burden shifts to the Secretary to show that the claimant can engage in other types of substantial gainful work that exists in the national economy.
Id.
In this case, the
ALJ
found that Andrews’s psoriasis prevented him from returning to his past relevant work, but found that Andrews had no mental impairments that, singly or in combination, were “medically equal” to those listed in 20 C.F.R. Pt. 404, Subpt. P., App. 1 §, 12.00.
Ill
Andrews offers a number of arguments concerning why the Secretary erred in denying benefits, but they all turn on whether the ALJ properly rejected the conclusions of the examining psychologist.
The magistrate judge upheld the ALJ’s rejection of McCono-chie’s conclusions that Andrews is severely limited in his work-related-mental abilities on two grounds: first, that McConochie’s opinion as a nontreating physician is not entitled to any particular deference; and second, that the ALJ’s decision is supported by substantial evidence as (1) MeConochie’s conclusions are based only on information from Andrews and Andrews isn’t credible, and (2) the testifying medical advisor’s conclusions are supported by opinions of the state disability reviewers under contract with the Secretary. Andrews takes issue with both aspects of the magistrate judge’s reasoning. He submits that McConochie’s opinion was tantamount to that of a treating source because he clinically examined Andrews four times in thirteen months. In addition, he argues, Green never examined him, and her and the ALJ’s reliance on reports of nonexamining medical ad-visors, none of whom made independent clinical findings, is insufficient to make Green’s opinion substantial evidence upon which the denial of benefits may be upheld. For this reason, Andrews contends, MeConochie’s opinion could only be rejected for clear and convincing reasons, which, he contends, the ALJ failed to give.
The Secretary emphasizes that there was no treating physician here, and that Andrews never sought treatment for the mental condition's upon which his claim is based. Further, the Secretary notes that an ALJ may properly consider a claimant’s lack of credibility and the extent to which his physician’s opinion is influenced by the claimant’s own information.
Fair v. Bowen,
885 F.2d 597, 604 (9th Cir.1989). Finally, she contends that because the reports of. medical advisors may serve as substantial evidence, Green’s testimony was consistent with reports of four other medical advisors, Andrews’s representations weren’t credible and McConochie was not a treating-physician, McConochie’s opinion was properly rejected for specific, legitimate reasons based on substantial evidence in the record.
We have frequently considered the law governing the opinions of treating physicians in disability cases, most comprehensively in
Magallanes.
However, we have not directly addressed the extent to which the
Magal-lanes
guidelines apply to the opinions of examining physicians and their relationship to the opinions of nonexamining medical advis-ors called by the Secretary. We must do so here, because there is no treating psychologist in this ease. Rather, the record consists of testimony and reports by an examining psychologist whom Andrews saw for clinical evaluation to get SSI benefits, reports of four medical/psychological advisors, and the testimony of the claimant and a nonexamining psychologist expert called by the ALJ.
A
It is clear that more weight is given to a treating physician’s opinion than to the
opinion of a nontreating physician because a treating physician “is employed to cure and has a greater opportunity to know and observe the patient as an individual.”
Magallanes,
881 F.2d at 751 (quoting
Sprague v. Bowen,
812 F.2d 1226, 1230 (9th Cir.1987)). Likewise, greater weight is accorded to the opinion of an examining physician than a non-examining physician. 20 C.F.R. § 416.927(d)(1);
see Pitzer v. Sullivan,
908 F.2d 502, 506 n. 4 (9th Cir.1990).
While the ALJ may disregard the opinion of a treating physician, whether or not controverted, the ALJ may reject an
uncontroverted
opinion of a treating physician only for clear and convincing, reasons.
Magallanes,
881 F.2d at 751. We have said that the same rule applies to the opinions of an examining physician in the absence of legitimate conflicting testimony and any reason for the ALJ’s rejection of the examining physician’s opinion.
See Pitzer,
908 F.2d at 506. Where the opinion of the claimant’s treating physician is contradicted, and the opinion of a nontreating source is based on independent clinical findings that differ from those of the treating physician, the opinion of the nontreating source may itself be substantial evidence; it is then solely the province of the ALJ to resolve the conflict.
Magallanes,
881 F.2d at 751. Where, on the other hand, a nontreating source’s opinion contradicts that of the treating physician but is not based on independent clinical findings, or rests on clinical findings also considered by the treating physician, the opinion of the treating physician may be rejected only if the ALJ gives specific, legitimate reasons for doing so that are based on substantial evidence in the record.
Id.
at 751, 755.
See Ramirez v. Shalala,
8 F.3d 1449, 1453 (9th Cir.1993) (applying test where ALJ relied on contradictory opinion of nonexamining medical advisor).
Andrews argues that the district court incorrectly relied on
Magallanes
to credit the opinions of the nonexamining hearing expert as substantial evidence supporting the ALJ’s finding of nondisability because Green’s opinions were not supported by the findings or opinions of an examining or treating expert and were not based on independent clinical findings. We agree with Andrews that the district court erred in stating that the examining physician’s opinion is not entitled to any particular deference. Both the regulations, 20 C.F.R. § 416.927(d)(1), and our precedent,
see Pitzer,
908 F.2d at 506 n. 4, state that the conclusion of a nonexamining expert is generally entitled to less weight than the conclusion of an examining physician. However, giving the examining physician’s opinion
more
weight than the nonexamining expert’s opinion does not mean that the opinions of nonexamining sources and medical advisors are entitled to
no
weight.
As we explained in
Magallanes,
where the ALJ relied on a nonexamining source’s testimony on limitations to reject the opinion of the claimant’s treating physician, the report of a nonexamining, nontreating physician need not be discounted when it “is not contradicted by
all other evidence
in the record.”
Id.
at 752 (emphasis in original). Reports of consultative physicians called in by the Secretary may serve as substantial evidence. Indeed,
[t]he analysis and opinion of an expert selected by the ALJ may be helpful to the ALJ’s adjudication, and we should not impose “burdensome procedural requirements that facilitate ... second-guessing [the ALJ’s resolution of conflicting medical testimony].”
Id.
at 753 (quoting
Allen v. Heckler,
749 F.2d 577, 580 (9th Cir.1984)). A fortiori, when it is an
examining
physician’s opinion that the ALJ has rejected in reliance on the testimony of a nonexamining advisor, reports of the nonexamining advisor need not be discounted and may serve as substantial evidence when they are supported by other evidence in the record and are consistent with it. We therefore decline Andrews’s invitation to hold that a nonexamining medical advisor’s statements inevitably deserve “little” or no weight.
Nor do we share Andrews’s view that the ALJ could only meet the burden for rejecting his examining psychologist’s opinion by giving clear and convincing reasons. In the eases upon which he relies that impose this burden,
Fife v. Heckler,
767 F.2d 1427 (9th Cir.1985),
Montijo v. Secretary,
729 F.2d 599 (9th Cir.1984), and
Pitzer,
there was no
evidence contrary to the opinion of the claimant’s physician — and conversely, none supporting the conflicting opinion of the ALJ or the Secretary’s consultant. In
Fife,
a speculative comment that the claimant might be able to return to work after removal of a cast did not contradict clinical, post-cast determinations of disability. In
Montijo,
there was no medical evidence or opinion contrary to that of the claimant’s psychiatrist and psychologist; indeed, there is no indication that the record contained any opinion of a medical advisor. And in
Pitzer,
there was nothing in the record to support the nonexamining physicians’ written reports and the Secretary’s conclusion that the claimant, who had not worked in fifteen years and couldn’t return to her former work, had the functional capacity to perform gainful work. For that reason, the nonexamining physicians’ opinions “with nothing more” could not constitute substantial evidence.
Pitzer,
908 F.2d at 506 n. 4. Here, there is legitimate conflicting testimony by Dr. Green and Andrews, and the ALJ gave specific reasons for disregarding the conflicting observations, opinions and conclusions of McConochie. We are thus not obliged to inquire whether the ALJ failed to give, clear and convincing reasons for rejecting McConochie’s opinions.
B
Our review of the record persuades us that the ALJ met the appropriate burden for discrediting McConochie’s diagnosis and rejecting his opinion that Andrews was severely handicapped by chronic drug abuse, depression and paranoid ideation such that he was a poor occupational candidate. Andrews consulted McConochie only because he had to in order to obtain benefits; McCono-chie himself observed that Andrews had no interest in pursuing treatment for his self-reported psychological symptoms.
Andrews admitted that he manipulated situations to his advantage, and that he self-treated his anxiety by smoking marijuana. Green, the nonexamining psychologist, testified as well. Because she was subject to cross-examination, the ALJ could legitimately credit her testimony.
See Torres v. Secretary of H.H.S.,
870 F.2d 742, 744 (1st Cir.1989) (greater weight may be given to opinion of nonexamining expert who testifies at hearing subject to cross-examination);
see also Ramirez,
8 F.3d at 1453 (noting that nonexam-ining physician testified at the hearing as a medical expert). Green had special expertise in the treatment of substance abuse; there is nothing to suggest that McConochie had any. Under the regulations, the ALJ could properly credit Green’s opinion on account of her expertise. 20 C.F.R. § 416.927(d)(5).
She stated that McConochie’s ability to diagnose Andrews accurately was compromised by Andrews’s substance abuse (the existence of which was undisputed) and that the reliability of MeConoehie’s diagnoses was further undermined because they were based entirely on “self report” by Andrews. Green pointed to Andrews’s inconsistent stories about his history of substance abuse as typical of substance abusers, and opined that the uncorroborated self report of substance abusers is unreliable. It was Green’s opinion, based upon Andrews’s testimony and medical history, that his condition was controllable. Finally, Green adopted Dr. Smith’s MRFC assessment that Andrews was no more than moderately, limited in any of the 20 standard categories. Green’s opinions were consistent with Andrews’s testimony, as well as with the written reports of the four other experts who reviewed Andrews’s medical records. She parted company with McConochie only insofar as he failed adequately to consider the diagnostic implications of Andrews’s substance abuse, an area in which she'possesses particular expertise.
The ALJ set all of this out in a detailed and thorough summary, including his interpretation of the evidence and a comparison with McConochie’s diagnosis and opinions.
Magallanes,
881 F.2d at 751. He found that MeConochie’s conclusions regarding depression, post-traumatic stress disorder and schizotypal personality disorder were unreliable because of Andrews’s contemporaneous substance abuse. The ALJ found that Andrews lacked credibility, based in part on the admission that he had learned to “manipulate” programs and people and “would play stupid” in court to avoid punishment, and based in part on the fact that Andrews’s disabilities appeared to cause “no significant disruption of his daily activities or concentration ability.” Credibility determinations are the province of the ALJ.
Magallanes,
881 F.2d at 750. Because McCono-chie’s diagnoses were based on the self reporting of an unreliable person, the ALJ decided to accord them less weight. This he could legitimately do; an opinion of disability premised to a large extent upon the claimant’s own accounts of his symptoms and limitations may be disregarded, once those complaints, have themselves been properly discounted.
Flaten v. Secretary of Health & Human Services,
44 F.3d 1453, 1463-64 (9th Cir.1995). As the ALJ then stated, the question narrowed to whether Andrews was disabled by means of his inability to control the use of drugs or alcohol.
Tylitzki v. Shalala,
999 F.2d 1411, 1413 (9th Cir.1993) (alcoholism is not per se disabling, but can constitute a disability if it is uncontrollable). In this regard, the ALJ pointed to Andrews’s testimony that he had been able to control his alcohol and drug dependence for years at a time, to Green’s conclusion that Andrews was not an uncontrollable substance abuser, and to medical reports in the record indicating that Andrews had been able to stop his drug and alcohol use for extended periods of time.
In sum, the ALJ gave specific, legitimate reasons for rejecting the examining psychologist’s opinions and for relying on the nonex-amining medical expert’s opinion instead, which was based on a review of the medical records, was consistent with the written reports of four other experts and with Andrews’s testimony, and legitimately controverted McConochie’s observations, opinions and conclusions. The ALJ’s reasons were based on Andrews’s testimony, weaknesses in the examining psychologist’s diagnosis and conclusions as shown by Andrews’s testimony and Green’s opinion, the opinions and analysis of the nonexamining psychologist who testified as a medical expert, and the written reports of other medical advisors. This constitutes substantial evidence. We therefore hold that the ALJ met his burden of giving specific, legitimate reasons based on substantial evidence for rejecting the examining physician’s opinion on mental residual functional capacity in favor of the nonexamining expert’s opinion.
IV
Andrews next argues that, even if the Secretary properly chose Green’s residual functional capacity assessment over McCono-chie’s, she has failed to carry her burden of showing that Andrews is capable of performing specific jobs in the economy because the hypothetical posed to the vocational expert did not reflect all of Andrews’s functional limitations as found by Green. We agree.
If a claimant shows that he cannot return to his previous work, the Secretary must show that the claimant can do other kinds of work.
Magallanes,
881 F.2d at 756. The Secretary may carry this burden by eliciting the testimony of a vocational expert in response to a hypothetical that sets out all the limitations and restrictions of the claimant.
Id.
Although the hypothetical may be based on evidence which is disputed, the assumptions in the hypothetical must be supported by the record.
Gallant,
753 F.2d at 1456. As we have held, Green’s assessment of Andrews was supported by the record.
Andrews, however, argues that the hypothetical put to the vocational expert, Vernon Arne, did not adequately reflect even Green’s residual functional capacity assessment. Green adopted Dr. Smith’s assessment, which found Andrews “not significantly limited” or “without evidence of limitation” in 14 of 20 categories. Smith rated Andrews “moderately limited” in his ability to interact appropriately with the general public, and in
three “concentration and persistence” categories: ability to understand and remember detailed instructions, to carry out detailed instructions, and to work in coordination with or proximity to others without being distracted by them. Andrews was also rated moderately limited in two “adaptation” categories: ability to respond appropriately to changes in the work setting and to set realistic goals or make plans independently of others. According to Green, Andrews is markedly limited in no category.
The relevant part of the ALJ’s hypothetical stated:
I do want you to assume that ... we ... have an individual who should have an occupation that involves minimal'interaction. Weren’t frequent [sic] dealing with the public and coworkers.... [W]ould the carpenter’s labor be able to be performed under those restrictions.
The ALJ’s hypothetical refers only to Andrews’s social limitations and not to his moderate limitations in other areas. Arguably, Arne clarified the ALJ’s hypothetical in connection with Andrews’s restricted ability to follow instructions by responding that, as a carpenter, “you do have to receive instructions from a supervisor” and that it “isn’t a position you can perform in an isolated manner.” Apparently, the idea of carpentry was dropped; Arne then proposed the jobs of machine off-bearer, hardware assembler, hand packager, and wire preparer as jobs that' would “fit your hypothetical as involving no public contact and probably minimal interaction with coworkers.” However, there is no indication that Arne considered Andrews’s limitations in “adaptation.” Since the hypothetical must consider all of the claimant’s limitations,
Magallanes,
881 F.2d at 756, the ALJ’s hypothetical was insufficient to carry the Secretary’s burden of proving ability to engage in work in the national economy. Arne was present throughout the hearing, but we cannot assume that he ventured beyond the ALJ’s hypothetical and incorporated all limitations discussed by Green in her testimony. Accordingly, we reverse and remand this ease to the district court with instructions to remand to the Secretary for a determination of vocational ability based upon a hypothetical that accurately reflects Andrews’s mental residual functional capacity-
REVERSED and REMANDED.