OPINION OF THE COURT
BECKER, Circuit Judge.
Under the Black Lung Benefits Act, disability claims ultimately come before an Administrative Law Judge (ALJ) who must, inter alia, determine the reliability of medical evidence submitted to prove that the claimant is totally disabled. The regulations promulgated by the Secretary of Labor provide that the AU may find a miner totally disabled on the basis of a physician’s judgment if the judgment is “reasoned” and based on medically acceptable evidence. 20 C.F.R. § 718.204(c). In this case, we must decide whether the Secretary has limited the AU’s discretion under § 718.204(c) by providing that a judgment contained in a medical report “alone” may support a finding of total disability only if the report is in “substantial compliance” with a quality standard set out at 20 C.F.R. § 718.104. That section mandates that a report include a medical and employment history, describe certain test results and certain symptoms. The Secretary’s designate, the Director of the Office of Workers’ Compensation Programs (OWCP) advocates this view.
The Benefits Review Board (BRB) rejected the Director’s contention. Following one of its prior decisions based both on construction of the regulations and statutory grounds, the BRB held that the regulation’s quality standards, of which the medical report standard is only one, do not preclude AU’s from relying on noncomplying evidence. In the BRB’s view, the quality standards have a mandatory effect only on the OWCP’s internal decisionmaking not on ALJ’s, who adjudicate claims only after the OWCP has denied them. The BRB therefore affirmed an ALJ’s decision granting benefits to respondent August Mangifest despite the ALJ’s primary reliance on a medical judgment contained in an apparently noncomplying report.
The Director now petitions for review. On the basis of our own interpretation of the regulations, but for different reasons than those of the BRB, we agree with the BRB that an AU may find a miner totally disabled in reliance upon a medical judgment in a noncomplying report so long as the judgment is reasoned and based on medically acceptable evidence as required by § 718.204(c)(4). We believe that the ALJ must make this decision “in accordance with” § 718.104, but only in the sense that that section should serve as a guide: it identifies the kinds of information AU’s should expect a physician to rely on in finding total disability, and it instructs the hearing officer to examine the evidence in determining whether the physician’s diagnosis was reasoned. Such a rule is consistent with the Black Lung statutes and the APA.
We ground our decision on “our own interpretation” notwithstanding the deference normally required to an interpretation of the Director because we find ambiguities and inconsistencies in the Director’s interpretation of the regulations that are sufficiently great to preclude deference in this [1320]*1320case. However, we also believe that our result is ultimately consistent with the views of the Director: the test the Director suggests for evaluating a report’s substantial compliance with the quality standards of § 718.104 requires the AU to engage in the same analysis of the documentation and reasoning of the report’s medical judgment that we find required by § 718.-204(c)(4).
Although we reject a rigid approach to the quality standard regulation, we grant the petition for review and remand this case for further proceedings. The ALJ did not use the quality standard as a guideline in determining whether the medical judgments he relied on were documented and reasoned. The ALJ also failed specifically to find the medical judgment documented and reasoned or to explain that finding. Clear articulation was necessary because the medical report apparently deviated substantially from the quality standard.
I. Regulatory Structure
Disability claims filed after March 31, 1980 are evaluated under the criteria established by Part 718 of the permanent black lung regulations. See 20 C.F.R. § 718.2, 725.4(a) (1987). Under Part 718, the miner must prove that he has pneumoconiosis, that he contracted it through his coal mine employment, and that he is totally disabled due to the disease. See 20 C.F.R. §§ 718.-201-.204 (1987).
Subpart C of the Part 718 regulations, 20 C.F.R. §§ 718.201-.206, establishes criteria for finding these three elements necessary to a black lung claim. At issue in this case are the criteria for finding total disability set out in § 718.204. Subsection (b) of that section defines total disability: a miner is totally disabled if he is incapable of performing his usual coal mine work and if he cannot engage in gainful employment in the immediate area of his residence requiring skills comparable to those he used in his mine employment. See also 30 U.S.C. § 902(f)(1)(A). Subsection 718.204(c) then sets out medical criteria for determining whether a miner meets this standard. It provides that an AU must find total disability, in the absence of contrary evidence, if pulmonary function or arterial blood gas tests produce certain results, § 718.-204(c)(l-2), or if the medical evidence demonstrates that the miner has cor pulmonale with right sided congestive heart failure, § 718.204(c)(3). In addition, paragraph 718.204(c)(4),1 the paragraph most directly relevant to this case, provides that an AU “may” find total disability if a physician “exercising reasoned medical judgment, based on medically acceptable clinical and laboratory diagnostic techniques concludes that a miner’s respiratory or pulmonary conditions prevents or prevented the miner from engaging in employment” as defined by subsection (b).2
[1321]*1321This case concerns the relationship between the criteria in § 718.204(c)(4) and § 718.104 in subpart B of the Part 718 regulations. Subpart B generally sets out quality standards for most kinds of medical evidence relevant to Black Lung claims, including X-rays, pulmonary function studies, arterial blood-gas studies, and autopsies and biopsies.3 Section 718.104, which we set out fully below,4 provides the standard for medical reports. It requires that a report “of any physician examination conducted in connection with a claim” use either a form supplied by the OWCP or contain “substantially the same information.” The information required includes the miner’s medical and employment history, the results of medical tests, and all manifestations and symptoms of chronic respiratory or heart disease.
In Budash v. Bethlehem Mines Corp, 9 Black Lung Rep. 1-48 (en banc), aff'd on reconsideration en banc, 9 Black Lung Rep. 1-104 (1986), the BRB considered the question whether a medical report not in compliance with the quality standard at § 718.104 may nevertheless justify a finding of total disability under § 718.204(c)(4). Rejecting the Director’s contention, the BRB held that it could. The BRB did not dispute the authority of the Labor Secretary generally to establish criteria and other reasonable rules that bind the AU’s consideration of claims. See 30 U.S.C. § 902(f)(1), § 936. The BRB believed, however, that the exclusion of non-complying reports would violate the statutory requirement set out at 30 U.S.C. § 923(b) that claims decisions be based on “all relevant evidence.” The BRB also believed that section 7 of the Administrative Procedure Act, 5 U.S.C. § 556(c), requires consideration of all relevant evidence and required rejection of the Director’s position. Finally the BRB felt that the regulations themselves do not require the exclusion of non-complying evidence.
The BRB accordingly held that § 718.104 has a mandatory effect only on the OWCP’s internal decision whether to grant or deny claims.5 While ALJ’s “may” consider the quality standard, the BRB believed, those standards are not binding upon them. 9 Black Lung Rep. at 1-50 to 1-51.6
II. Facts and Procedural History
Respondent August Mangifest is a seventy-two year old resident of Vandergrift in Western Pennsylvania. From 1929 until 1941, Mangifest worked in underground coal mines as a loader, driller and weigh-man. Following this coal mine experience, [1322]*1322Mangifest worked in a váriety of jobs, most recently as a part-time janitor and as a restaurant supply salesman. On April 8, 1980, Mangifest filed a claim for benefits under the Black Lung Benefits Act as amended, 30 U.S.C. §§ 901-45. (1982). The OWCP denied the claim and Mangifest requested a formal hearing before an AU.
All parties agree that Mangifest has penumoconiosis and that he contracted it through his employment as a coal miner. The Director of OWCP and Mangifest dispute, however, whether Mangifest is “totally disabled.” Because the Director offered no information to the AU to show that Mangifest is not disabled, the dispute pertains entirely to the adequacy of Mangifest’s proofs.
Taken alone, neither Mangifest’s pulmonary functions studies nor arterial blood-gas tests established total disability under §§ 718.204(c)(1) or (2). Nor does Mangifest suffer from cor pulmonale, which may establish total disability under § 718.-204(c)(3). To prove total disability, Mangifest therefore submitted the reports of four physicians. Although all agreed that Mangifest suffered from pneumoconiosis, the reports differed in their comprehensiveness and in their conclusions bearing on Mangifest’s degree of disability.
Mangifest submitted two reports which carefully listed symptoms and test results, one from Dr. Arthur E. Barnes and one from Dr. J.D. Silverman. Dr. Barnes’s report offered no assessment of the degree of disability. Dr. Silverman’s report concluded that Mangifest suffered from anthracosilicosis and “should not be permitted to return to work in any dusty atmosphere.” (A. at 106) The Director contends that neither of these reports can establish total disability because neither found total disability. According to the Director, Dr. Silverman’s finding that Mangifest “should not be permitted to return” to a dusty environment was not a finding that Mangifest is totally disabled.
In addition, Mangifest submitted reports from two members of the Russellton Medical Group clinic. Dr. Thomas B. Connelly’s report stated that Mangifest “is totally and permanently disabled” and that he “would be unable to do coal mining work.” But Dr. Connelly’s report was a mere one page letter. It listed no medical results other than Mangifest’s statement of symptoms and an X-ray.7 Dr. Jeffrey M. Wolff, found by the AU to be Mangifest’s treating physician, provided a similar one page report. It stated that Mangifest is “totally and permanently disabled” and “is unable to do any further coal mining work.” Although it listed more symptoms and test results than Dr. Connelly’s letter, the report failed to mention ventilatory function studies or several other pieces of information required by the form.
The Director claims that neither Dr. Connelly’s nor Dr. Wolff’s report may support a finding of total disability because neither was in “substantial compliance” with § 718.104. Because of their brevity, there seems little dispute that the reports in and of themselves fail to satisfy the requirements of § 718.104. Despite the facial inadequacy of their reports, however, both Drs. Connelly and Wolff may have based their judgments on the wide range of evidence required by the medical report standard. The record establishes that Mangifest had been seen at the Russellton clinic for over one year, and that objective tests, medical histories, and other examinations had been conducted at the clinic. Mangifest submitted many of those results to the AU separately.
Even if the AU may consider the reports of Drs. Connelly and Wolff, the Director argues that no finding of total disability is permissible because the reports are not reasoned. The Director contends that the doctors’ conclusions of total disability were based entirely on evidence that does not measure the degree of disability.
The AU disagreed with the Director. He held that Dr. Wolff's report did not [1323]*1323have to meet the quality standards of § 718.104 because the fact that Dr. Wolff was Mangifest’s treating physician meant that the report was not prepared in connection with the claim for benefits. According to the ALJ, the report therefore could be a basis for finding total disability under § 718.204(c)(4). The AU found the report credible because it was partially corroborated by the opinions of the other doctors. The ALJ concluded that Mangifest was totally disabled due to pneumoconiosis and awarded benefits from April 8, 1980, the date Mangifest filed his application.
The Director appealed to the BRB. The BRB assumed for purposes of its consideration that Dr. Wolff’s report was prepared in connection with the claim. The BRB further assumed that the report was not in substantial compliance with the quality standards set out at § 718.104. Relying on Budash, however, the BRB held that the Dr. Wolff’s report could support a finding of total disability under § 718.204(c)(4). In response to the Director’s claim that Wolff’s and Connelly’s medical reports were not reasoned, the BRB noted that such a judgment was a matter for the AU’s discretion. Finally, the BRB found that the AU had based his finding of total disability on “substantial evidence” and therefore affirmed.
In this court, the Director renews the arguments he made before the BRB.8 He stresses the need for deference to his position as the delegate of the Secretary of Labor. Three mining or insurance companies, appearing as amici curiae, support Mangifest’s contention that a noncomplying medical report can justify a finding of total disability.9 Mangifest, however, has abandoned any contention that Dr. Wolff’s report was not prepared in connection with a claim.10 Because of our construction of the regulations, we reach only some of the statutory issues presented to us.
III. The Standard for Dealing with Medical Reports
A. Deference
The Supreme Court has made clear that courts must defer to an agency’s consistent interpretation of its own regulation unless it is “plainly erroneous or inconsistent with the regulation.” Bowles v. Seminole Rock and Sand Co., 325 U.S. 410, 414, 65 S.Ct. 1215, 1217, 89 L.Ed. 1700 (1945); New Haven Bd. of Ed. v. Bell, 456 U.S. 512, 102 S.Ct. 1912, 72 L.Ed.2d 299 (1982); Revak v. National Mines Corp., 808 F.2d 996 (3d Cir.1986); see generally R. Weaver, Judicial Interpretation of Administrative Regulations: The Deference Rule, 45 U.Pitt.L.Rev. 587, 620-22 (1984). We owe such deference to the Director, not to the BRB, for the Director is the maker of policy. See Potomac & Electric Power Co. v. Director, OWCP, 449 U.S. 268, 278 n. 18, 101 S.Ct. 509, 514 n. 18, 66 L.Ed.2d 446 (1980) (requiring deference to Director not to BRB in matters of statutory interpretation); Bethlehem Mines Corp. v. Director, OWCP, 766 F.2d 128, 130 (3d Cir.1985) (requiring deference to Director not to BRB in construction of regulations).11
[1324]*1324We have cautioned, however, that this deference does not permit us to defer to an “interpretation” in an adversary proceeding that strains “the plain and natural meaning of words in a standard to alleviate an unlikely and uncontemplated hazard.” Bethlehem Steel v. OSHA, 573 F.2d 157, 161 (3d Cir.1978). As the Supreme Court in Bowles deferred to the agency’s explanation of what a particular phrase in a regulation meant, we defer to a policymaker’s plausible explanation of the language in a regulation. That rule, however, does not permit the policymaker, in an adjudicatory proceeding, to imply language that simply does not exist:
The responsibility to promulgate clear and unambiguous standards is upon the Secretary. The test is not what he might possibly have intended, but what he said. If the language is faulty, the Secretary has the means and the obligation to amend.
Bethlehem Steel, 573 F.2d at 161.
Furthermore, we have in the past differentiated between a reasoned interpretation of a regulation’s language and a mere position about what the regulations require. See Bevak, 808 F.2d at 1003 (refusing to defer to agency view in absence of “any explanation for how [court] might find support for the [agency’s] position in the language of the regulation”). Although we must defer to an agency regulation that is not “plainly erroneous,” we must understand how the agency connects its position to the language of the regulation in order to evaluate its plausibility.12 If we cannot understand the agency’s reasoning, if it is self-contradictory, or if it is ambiguous, we cannot defer to it.
In this case the Director contends that because “the reports of Drs. Wolff and Connelly are not in substantial compliance with [the] regulation, the AU could not rely upon either report in determining that Mangifest was totally disabled by pneumoconiosis.” Director’s Brief at 35. This statement in the Director’s brief is consistent with the views of the Director’s counsel at oral argument. She contended that an AU could not use noncomplying evidence even to tip the scales of a decision that had the support of complying evidence on both sides. In that view, non-complying evidence would be worthless because it could never make the difference between one result and another.
This severe view, however, is inconsistent with the interpretation of the regulations otherwise expressed in the Director’s briefs. At several points, the Director takes the position only that a non-complying medical report may not, “standing alone, sustain the party’s burden of proof on an element of entitlement.” Director’s Brief at 17. Apart from that one prohibition, such evidence “is to be assigned weight by the trier-of-fact based on all circumstances of the case.” Id.
In this case, the AU did not find total disability on the basis of Dr. Wolff’s report alone. Instead, he found total disability because he found Dr. Wolff’s report corroborated by the other three reports. This finding appears to be permissible under the less strict interpretation of the Director presented to us. The discrepancy between [1325]*1325the Director's less severe view of what the regulations require and alternative submissions as to what they mandate in this case makes us feel less than confident that we fully understand even his less severe interpretation.
Not only does the Director present conflicting interpretations, but he has difficulty connecting each view with the language of the regulations. As we discuss more fully below, the Director bases his analysis of § 718.104 on a comparison with other quality standards. Quality standards for ventilatory studies, for x-rays, and for arterial blood-gas studies state that no test can constitute evidence unless it is in substantial compliance with the quality standard. The Director contends that we should read § 718.104 as if it included similar language for medical reports.
If § 718.104 included such language, it would state that a non-complying report cannot be evidence at all. It would therefore support the severe view. For us to infer language that does not otherwise exist, however, we would have to to defer to the Director’s “subjective” view of the regulation, not to its objective meaning. A claimant proceeding in good faith should not be subjected to a trap brought about by an interpretation of a regulation hidden in the bosom of the agency.
Furthermore, even if § 718.104 contained language like that of the other quality standards, it would not support the lenient interpretation. The section would state that a noncomplying report cannot be evidence at all; it would not state that a non-complying report can be evidence when combined with other evidence but cannot be evidence when considered alone. The Director has therefore provided us with no way in which we can connect the more lenient interpretation with the language of the regulations.
Because of these inconsistencies in the Director’s views and because of the Director’s failure to provide a reasoned explanation of the regulations, we can find no coherent view of the way the regulations interact to which we owe deference. The Director’s understanding of some particular regulatory sections is clear and objectively supportable. We will defer to them as appropriate. For the way in which the regulations fit together however, we must rely on our own interpretation.
B. Interpretation of the Regulations
In its Budash decision, the BRB supplied two reasons based solely on the regulations to support its view that the medical report quality standard does not apply to AU deliberations except to the extent that the AU chooses to use them as a guide. First, it held that the quality standards in Part 718 subpart B generally apply only to the internal decisionmaking of the OWCP because they are entitled “Criteria for the Development of Medical Evidence.” Because the OWCP is responsible for the development of the medical evidence, the BRB reasoned, this title indicates the internal purpose of the guidelines.
We agree with the Director, however, that Part 718, Subpart B standards are not mere, internal agency guidelines. Section 725.456(c) of Title 20 C.F.R. provides explicitly that “ [a]ll medical records and reports submitted by any party shall be considered by the administrative law judge in accordance with the quality standards contained in Part 718 as amended from time to time.” Thus, the AU must make benefit decisions “in accordance with” the quality standards, including § 718.104.
Second, the BRB reasoned that if § 718.-104 controlled the AU’s findings, it would conflict with § 718.204(c)(4), which authorizes an AU to find total disability if a medical judgment is documented and reasoned. According to the Director, however, § 718.104 “insures no more than that the reports contain sufficient documentation upon which a physician may form a reasoned medical judgment.” Director’s Brief at 28-29. In contrast, § 718.204 contains the additional requirement that the judgment be “reasoned.” Id. We find the Director’s reasoning persuasive and disagree with the BRB that the two sections necessarily conflict. The alleged conflict between the two sections is no basis for [1326]*1326concluding that § 718.104 applies only to the OWCP and not to AU’s.
Although we therefore agree with the Director that the AU must make decisions “in accordance with” the quality standards, that requirement yields no clear command in this case. The quality standards for ventilatory function studies, X-rays, and autopsy or biopsy evidence provide explicitly that non-qualifying test results shall not constitute evidence unless they are in substantial compliance with the standards. 20 C.F.R. §§ 718.102(e), 718.103(c), 718.-106(b).13 In contrast, the § 718.104 standard for reports of physican examinations contains no such “substantial compliance” language. The essence of the Director’s argument is that the existence of the language of exclusion in three sections implies similar treatment of § 718.104. Unless we assume general sloppiness, however, the omission in § 718.104 suggests a consciously different policy.
The arterial blood-gas studies subsection also contains no “substantial compliance” language. Perhaps this omission suggests that the drafters were in fact sloppy and that the omission in the medical report standard may have been inadvertent. But sections of other Parts of the black lung regulations require an AU to consider noncomplying medical reports even while they exclude noncomplying medical tests.
For example, the regulations applicable to proceedings under the interim presumption of Part 727 require that X-ray, ventilatory study or blood gas study evidence that does not comply with the quality standards of Part 718 shall not be sufficient to invoke the presumption. 20 C.F.R. § 727.208 (1987); see also Strako v. Zeigler Coal Co., 3 Black Lung Rep. 1-136 (B.R.B.1981) (noncomplying ventilatory studies cannot invoke interim presumption). But medical reports are not included in this rule. The BRB has accordingly held noncomplying medical reports sufficient to invoke the interim presumption under § 727.203(a)(4). See Parsons v. Director, OWCP, 6 Black Lung Rep. 1-273 (B.R.B.1983).14 This example suggests a general agency policy of treating deviations from medical report standards more leniently than deviations from other standards.
Section 718.206 of the regulations also supports this view. That section states:
Decisions, statements, reports, opinions, or the like, of agencies, organizations, physicians or other individuals, about the existence, cause and extent of a miner’s disability, or the cause of the miner’s death, are admissable. If properly submitted, such evidence shall be considered and given the weight to which it is entitled as evidence under all the facts before the adjudication officer in the claim.
This section appears to delegate discretion to the AU to determine the weight to which a doctor’s opinion “is entitled” under all the facts of the case. Indeed, the Director agrees that an AU has broad discretion generally although he claims that one rigid rule restricts this discretion, i.e., that noncomplying evidence alone cannot demonstrate total disability.15 But § 718.206 does not differentiate between complying and noncomplying physicians’ reports. Neither does it differentiate between the use of a physician’s report alone and its use with other evidence. Instead, § 718.206 re[1327]*1327quires the AU to determine “under all the facts” what weight to accord any physician’s report. Nothing in the regulations themselves, as opposed to statements by the Director in this litigation, suggests to the contrary.
Common sense explains why the Secretary would treat noncomplying medical tests and noncomplying medical reports differently. A medical test administered incorrectly may very well be completely unreliable. The Secretary so indicated in comments attached to the original publication of the regulation.16
The report of a physician about a miner’s degree of disability, however, may have a great deal of significance even if a report lacks full documentation. The report does not necessarily indicate the information upon which the physician relied. It is often buttressed by deposition testimony. In addition, a medical report carries with it the unspoken but powerful ally of the doctor’s professional judgment. Like other judgments, a medical judgment is sometimes based upon instinct, the unarticulated and unarticulable opinion that is nonetheless grounded in years of experience. Apparently out of respect for this medical intuition, the regulations permit an AU to find total disability on the basis of a medical judgments even if the medical tests are inconclusive.
Further explaining why § 718.104 might not require the exclusion of noncomplying medical reports, the Director’s own contentions to us in this case indicate that not all the medical requirements of § 718.104 relate to all of the facts relevant to a disputed claim. For example, the Director informs us that an X-ray is not normally relevant to the degree of disability. If the physician’s report fails to mention an X-ray, therefore, that failing should not normally affect the credibility of the physician’s finding of total disability.
Finally, the statutory language suggests a distinction between medical reports and medical tests. Code section 902(f)(1)(D) of the Black Lung Act obligates the Secretary to establish criteria to assure the accuracy of medical tests. See supra note 10. Whether the Secretary may or may not find in this section the authority to establish criteria for medical reports, see infra note 23, Congress’ concern for reliability obviously focused most acutely on tests, not reports.
For all these reasons, we do not construe the regulations to require the exclusion from an AU’s consideration of noncomplying medical reports. Instead, we hold that a medical judgment contained in a noncomplying report may constitute substantial evidence of total disability if, as required by § 718.204(c), it is “reasoned” and “based on medically acceptable clinical and laboratory diagnostic techniques.” In accordance with § 718.206, the AU must base this determination on all the facts of the case.
Section 718.104, however, is obviously not meaningless, and § 725.456(c) requires that an AU make a decision “in accordance with” it. It sets forth guidance about what information a physician should normally consider in making medical judgments. It also reflects an intention to have the AU examine the evidence on which the physician relied. In determining whether a report is reasoned and documented under § 718.204(c), the AU must therefore look to § 718.104 for guidance.
C. The Apparent Congruence Between the Director’s Interpretation and Our Own
Under our construction of the regulations, as we have just explained, we do not read into § 718.104 language to the effect that a report not in substantial compliance cannot constitute evidence. Even were we to accept the Director’s contention that we should do so, however, we believe that the ultimate requirements of the regulations applicable to this case would not change. As presented by the Secretary, the analysis [1328]*1328an AU must use to decide whether a report is in “substantial compliance” with § 718.104 appears the same as the analysis the AU must use under § 718.204(c)(4) to determine if a medical judgment is “reasoned” and “based on medically acceptable clinical and laboratory diagnostic techniques.”
The Director certainly appears to be of the view that § 718.104 and § 718.204(c)(4) require different analyses. We do not contend that they necessarily coalesce. Nor do we contend, as Amici suggest, that a substantial compliance standard is hopelessly vague. The question whether a report is in “substantial compliance” could focus entirely on the adequacy of the report, not taking into account the other facts of the particular case that might influence the reliability of the judgment expressed in the report. The AU would determine only whether the report lacked a piece of information that normally bears a significant relation to a judgment’s reliability.
For example, if a report lacked a date, as required by the form provided pursuant to § 718.104, the AU might determine that this technical deficiency did not make the report unreliable. On the other hand, if the report lacked the results of an X-ray, which normally bears a significant relationship to its reliability, the AU would certainly determine that the report is not in “substantial compliance.”
Using this analysis, an AU might determine that a report is not in substantial compliance even if the judgment expressed in it would satisfy the requirement of § 718.204(c) that it be documented and reasoned. For example, the report lacking an X-ray would presumably be noncomplying but it would still be sufficiently documented and reasoned to support a medical judgment that the claimant was totally disabled. (Perhaps such a rule would exist to encourage complete reports that make the determination of reliability easier.)
The Director, however, has disclaimed this view of the substantial compliance test. He states in his brief that “ ‘substantial compliance’ determinations must be made on a case-by-case basis, taking into account all relevant facts in the record” and that “it does not serve to exclude credible, probative evidence.” Director’s Reply Brief at 13-14. The Director stresses that the “substantial compliance” standard “is a rule of reason, a dynamic concept, governing discretionary determinations by the factfinder and it therefore cannot be stated in precise terms. To do so would be to take away the very discretion accorded the trier-of-fact under the regulations and the statute.” Director’s Reply Brief at 13.17
In oral arguments before the BRB in the case of Hucker v. Consolidated Coal Co., 9 B.L.R. 1-137 (1986) (en banc), counsel for the Director stressed, “(W]hat’s reasonable is what’s reliable.” Transcript provided in Appendix to Brief of Amici at 27F. Counsel also stated, “[Y]ou have to look at the issue for which the evidence is offered____” Id. at 27H.
Because the Director refuses to cabin the “substantial compliance” analysis, we find the Director’s understanding of the “substantial compliance” test to be indistinguishable from the requirement in § 718.-204(c) that a medical judgment be documented and reasoned. Both analyses require a discretionary determination by the AU based on all the facts of the case. The touchstones of both appear to be the reliability and documentation of the medical judgment expressed in the report not on the general reliability of the report. Accordingly, under the Director’s analysis, an AU would determine that a report was not in substantial compliance and thus exclude it only if the medical judgment was also not documented and reasoned and thus not capable of establishing total disability under § 718.204(c)(4).
[1329]*1329Following the Director’s analysis, we therefore return to the same regulatory rule that we reached without any consideration of agency deference. Even according to the Director: (1) the AU may credit a judgment in a medical report if it is documented and reasoned; (2) the AU has broad discretion to make this determination on all the facts of a case; and (3) § 718.104 operates essentially as a guide. Despite our disagreement with the path the Director follows in reaching this rule, we therefore feel reasonably confident that we implement the same result.
We recognize that we may not fully comprehend the Director’s views and that the Secretary may wish to clarify the regulations to express his policy more clearly. We trust that the Secretary, if he considers the matter, will not propose a rigid rule which will require the AU to discredit helpful medical evidence solely in the interest of administrative convenience but rather will attempt to strike a balance between the goals of administrability and reliability.
We note also that the gap between form and substance may be bridged if the OWCP, pursuant to its obligation to develop the record, see 20 C.F.R. § 725.401 (1987),18 attempts to discover the documentation and reasoning behind noncomplying reports. When Congress passed the portion of 30 U.S.C. § 902(f)(1)(D) which mandates that decisions be made on the basis of all the relevant evidence, the Senate report indicated Congress’ desire that “evidence be sought” by the government “as is necessary to assure a decision on the claim consistent with the remedial nature of this legislation.” Senate Report, supra note 5, reprinted at 2322. The report specifically pointed to the importance of evidence of the “miner’s physician.” Id., reprinted at 2318.
Rather than simply discrediting a medical report that does not contain sufficient documentation, the OWCP might therefore be advised to investigate the doctor’s reasons for finding total disability.19 By presenting that evidence to the AU, the OWCP could help insure that hearings concentrate on the reliability of doctor’s judgments not on the technical sufficiency of the doctor’s report.
IV. Statutory Requirements
Before applying our construction of the regulations to the facts of this case, we must consider Mangifest’s statutory arguments. With the support of Amici and the BRB, Mangifest has claimed that a “mandatory” construction of this regulation would violate 30 U.S.C. § 923(b) of the Black Lung Act and 5 U.S.C. § 556 of the APA. Under our middle-ground construction of the regulations, § 718.104 is mandatory (although not rigid): an AU may accept a medical judgment in a report that does not comply with § 718.104, but an AU must determine the reliability of that judgment with reference to the documentation requirements of that section.
We are not certain whether Mangifest’s rejection of a “mandatory” quality standard applies to the kind of middle-ground construction of the regulations we have accepted. The arguments in Mangifest’s and the Amici’s briefs and in the BRB’s decision focus primarily on the Director’s more rigid construction of § 718.104. Their reasoning is sufficiently unclear that we cannot be sure whether Mangifest, Amici and the BRB reject any mandatory quality standard or only a rigid one.20 Be[1330]*1330cause these briefs and the BRB opinion do claim to reject a “mandatory” regulation, however, and because § 718.104 is mandatory in our view, we believe we must consider whether our construction of the regulations violates the Black Lung Act or the APA by fettering the AU’s discretion.21
Code section 923(b) of the Black Lung Act states in pertinent part:
In determining the validity of claims under this part, all relevant evidence shall be considered, including, where relevant, medical tests such as blood gas studies, X-ray examination, electrocardiogram, pulmonary function studies, or physical performance tests, and any medical history, evidence submitted by the claimant’s physicians, or his wife’s affidavits____ [emphasis supplied]
Congress included this language in the Black Lung Benefits Act of 1972, Pub.L.No. 92-303, § 4(f), 86 Stat. 150, 154 (1972). As the legislative history reveals, Congress’s primary concern was the administrative practice of rejecting claims entirely on the basis of a negative X-ray. See Senate Rep. No. 92-743, 92d Cong., 2d Sess., reprinted in (1972) U.S.Code Cong. & Ad.News 2305, 2313-14. Experts testified that this reliance was inappropriate, id., and the primary purpose of the amendment was the prohibition against denying a claim solely on the basis of a negative X-ray.
The “relevant evidence” language in the amendment arose out of testimony in connection with the testimony involving the X-ray. Describing the “relevant evidence” provision, the Senate Report states:
In connection with testimony on the use of the X-ray as the sole basis for denial of a claim, witnesses throughout the hearing urged that provisions be made for more comprehensive evaluation and testing of the miner’s physical condition as it relates to respiratory and pulmo[1331]*1331nary impairments. This provision, taken in conjunction with the X-ray provision, would help achieve that objective, without strictly enumerating which tests shall be used or defining the limitations or extent of such tests.
Senate Report, supra, reprinted at 2317.
The origin of the “relevant evidence” language in Congress’ concern over the obsessive reliance on X-rays indicates that its primary purpose, as the Senate Report later stated, was “to expand the number of medical tools available” to the miner’s successful pursuit of his claim. Id. The statutory language itself reveals this purpose by enumerating the different medical tests Congress wanted the agency to consider.
Such a purpose is in no way thwarted by quality standards that mandatorily guide the AU’s consideration of the evidence. As the Senate Report stated explicitly, Congress expected the Secretary to adopt “evidentiary rules and disability evaluation criteria ... consistent with the language and intent of these amendments.” Id., reprinted at 2322 (emphasis supplied). We therefore believe that § 718.104 of the regulations does not violate the statute merely by establishing an evidentiary rule which the AU must use as a guide in determining claims.
We similarly find no problem in § 556 of the APA. That section discusses the use of evidence in several provisions. Subsection 556(c) provides:
Subject to published rules of the agency and within its powers, employees presiding at hearings may—
(3) rule on offers of proof and receive relevant evidence; [emphasis supplied]
Subsection 556(d) further provides inter alia:
Any oral or documentary evidence may be received, but the agency as a matter of policy shall provide for the exclusion of irrelevant, immaterial, or unduly repetitious evidence. A sanction may not be imposed or rule or order issued except on consideration of the whole record or those parts thereof cited by a party and supported by and in accordance with the reliable, probative, and substantial evidence____ [emphasis supplied]
Although § 556 does not command that agencies admit all relevant evidence, courts have often reversed agencies for excluding “facts and circumstances relevant to its inquiry which upon due consideration may be persuasive weight in the exercise of its discretion.” Atlas Copco, Inc. v. Environmental Protection Agency, 642 F.2d 458, 466-67 (D.C.Cir.1979).22 Courts have done so to insure parties “a fair opportunity to develop [their] position.” Second Tax Dist. of City of Norwalk v. F.E.R.C., 683 F.2d 477, 485 (D.C.Cir.1982). See generally K. Davis, Administrative Law § 16:11 (1979).
This concern for the exclusion of evidence, however, must be balanced against a concern for the use of unreliable evidence. While § 556(d) provides that an agency “may” receive relevant evidence, it also provides that no order may be issued that is not “supported by and in accordance with the reliable, probative, and substantial evidence.” Nothing in § 556 suggests that an AU must be free to make such a reliability determination without the guidance of regulations. To the extent that the APA mandates consideration of relevant evidence, we therefore do not believe that it [1332]*1332precludes an agency from establishing reasonable quality standards governing the admission of technical evidence. Cf. U.S. Dept. of Justice, Attorney General’s Manual on the Administrative Procedure Act (“nothing in section 7(c) [556] is intended to preclude an agency from imposing reasonable requirements as to how particular facts must be established”).
Neither § 923 of the Black Lung Act nor § 556 of the APA bar the Secretary from establishing quality standards that an AU must consider in determining the reliability of medical evidence. We therefore conclude that the regulations, as construed by us, are not in violation of any statutory requirements.
IV. Application of the Regulatory Standard to this Case
Even assuming that an AU can rely on nonconforming reports, the Director argues that we must reverse the BRB's order and deny benefits because neither Dr. Wolff’s nor Dr. Connelly’s reports indicate a documented and reasoned medical opinion. The Director argues that the factors relied on in those reports to indicate total disability do not measure disability, but at most are symptoms of some disability and indications of the disease.
The reports claim reliance on a variety of facts, however, including pulmonary symptoms and clinical examinations that are obviously relevant to a finding of total disability. In addition, evidence in the record from Dr. Wolff’s and Dr. Connelly’s Russellton clinic suggests that these doctors may have based their judgment on medical evidence that would have satisfied the formal requisites of § 718.104 if included in their reports. Their findings of total disability may or may not have been unreasonable, but we should not make the determination of reliability in the first instance.
Even if Dr. Wolff’s and the other reports clearly failed to satisfy the requirements of § 718.204(c)(4), we are not confident that these failures mandate rejection of Mangifest’s claim. The regulations do not make clear whether an AU may find total disability on the basis of criteria other than those established in § 718.204(c). The language of § 718.204(c) suggests that this section merely provides some criteria which, if satisfied, must result in such a finding. It states that “[i]n the absence of contrary probative evidence, evidence which meets the standards of either paragraphs (c)(1), (2), (3), (4) or (5) of this section shall establish a miner’s total disability.” (emphasis supplied)
Many other portions of the regulations suggest that an AU may find total disability on the basis of criteria not specified in § 718.204(c). Paragraph 718.204(e)(3), for example, states that some kinds of “changed circumstances of employment” may be “indicative of a miner’s reduced ability to perform his or her usual coal mine work.” Similarly, § 718.206, which we have discussed above, requires that “[decisions, statements, reports, opinions, or the like, of agencies, organizations, physicians or other individuals ... shall be considered and given the weight to which [they are] entitled as evidence under all the facts____” Furthermore, the Director has agreed that reports not prepared in connection with a claim and therefore not regulated by § 718.104 may support a finding of total disability under some circumstances. See § 718.107 (evidence not covered by quality standards “may be submitted in connection with a claim and shall be given appropriate consideration”).
If a wide range of factors may support a finding of total disability, then even the mere absence of a satisfactory report under § 718.204(c) may not doom Mangifest’s claim.
Although we cannot therefore state on this record that Mangifest does not deserve benefits, Dr. Wolff’s report, on which the AU relied pursuant to § 718.204(c)(4), obviously deviates considerably from the standards set out in § 718.104. Under our holding, these standards constitute mandatory guidelines by which the AU should have exercised his discretion. The AU, however, found § 718.104 completely inapplicable and thus apparently failed to use it [1333]*1333as a guideline for determining whether Dr. Wolffs report was documented and reasoned.
Furthermore, the AU failed explicitly to find Dr. Wolffs report, on which he chiefly relied, documented and reasoned. He accordingly failed to articulate the reasons for such a finding. Such an articulation was particularly necessary in light of the report’s considerable deviations from § 718.104. We therefore believe we must remand to the AU to assure that he indeed uses § 718.104 as a guide and to receive a fuller explanation of how § 718.204(c)(4) was satisfied.23
We will therefore grant the petition for review, reverse the decision of the BRB, and remand for reconsideration by the AU. Because the significance of § 718.-104 had not been clear and to insure that the ultimate decision reflects the weight of the evidence rather than technical considerations, Mangifest should be permitted to supplement his submissions to demonstrate why Dr. Wolff’s report is “reasoned” and “documented.”