J.T. Goodloe filed a claim for black lung benefits in 1978. Nineteen years later, and more than eight years after Goodloe’s death, the parties continue to grapple over Good-loe’s entitlement to benefits, and Goodloe’s attorney’s entitlement to fees. Unfortunately, we are unable to lend closure to this protracted case, but instead must remand because the administrative law judge applied an incorrect legal standard some twelve years ago.
BACKGROUND
An administrative law judge (“ALJ”) conducted a formal hearing on Goodloe’s claim on May 21, 1984, and denied the claim for benefits in a decision issued on May 6, 1985. The ALJ found that because Goodloe worked in coal mines for at least thirty-three years, he would be entitled to an interim presumption of total disability due to pneumoconiosis if he could meet any of the criteria set forth in 20 C.F.R. 727.203(a): (1) a chest x-ray, biopsy or autopsy established the existence of pneumoconiosis; (2) ventilatory studies established the presence of a chronic respiratory or pulmonary disease as demonstrated by certain values specified in the regulation; (3) blood gas studies demonstrated the presence of an impairment in the transfer of oxygen from the lung alveoli to the blood as indicated by certain values specified in the regulation; or (4) other medical evidence, including the documented opinion of a physician exercising reasoned medical judgment, established the presence of a totally disabling respiratory or pulmonary impairment. After finding that Goodloe did not meet the criteria set out in sections (a)(1) or (a)(2) because neither his chest x-rays nor his pulmonary function studies satisfied the requirements for the interim presumption, the ALJ determined that Goodloe did meet the standard set out in section (a)(3). Goodloe’s then-recent blood gas study produced qualifying values of PC02 equal to 37 and P02 equal to 45. That is, Goodloe’s blood contained so little oxygen that federal regulations presumed, because of his lengthy service in the mines, that he had developed pneumoconio-sis.
This blood gas study, conducted in 1984, differed so dramatically from a 1979 study of Goodloe’s blood that Peabody Coal Company, Goodloe’s employer, questioned its validity. Two doctors testified for Peabody and one testified for Goodloe, and all three viewed the 1984 blood gas results with suspicion. Good-loe’s doctor, Henry Peters, who ordered the 1984 blood test, testified that he considered the numbers odd. “I didn’t believe them myself when I first saw them, and so I had it double checked.” Deposition of Henry A. Peters, D.O., at 22-23. He called the hospi*209tal laboratory where the test was conducted to verify the results, and the laboratory confirmed to Dr. Peter’s satisfaction that this was in fact Goodloe’s blood gas reading. Although Dr. Peters could not remember how the laboratory verified the results — whether the lab repeated the test or verified the results in some other way — he repeatedly urged Peabody’s counsel to call the lab to cheek for himself. The record does not reflect whether Peabody’s counsel ever took Dr. Peters up on this suggestion.
Defense expert Dr: David W. Howard testified that, given Goodloe’s normal blood gas readings five years earlier, he would not accept a reading of P02 equal to 45 at face value, but would have the test repeated. Deposition of David W. Howard, M.D. at p. 6. He speculated that Goodloe’s blood had been drawn from a vein instead of an artery as it should be for a blood gas test, and that this error could account for the low reading. Id. at pp. 6-7. He further testified that his practice for patients with a verifiable P02 equal to 45 would be to put those patients on oxygen because this number indicated respiratory failure. Id. at p. 21.
The other defense expert, Dr. J. Frank Stewart, had examined Goodloe in 1979 when his blood gas readings were normal. Dr. Stewart testified that he thought he would ask for a repeat of the test if confronted with the “extremely unusually low” value of P02 equal to 45. “If I were treating that patient and got a report like that, compared to 91% at rest in 1979, five years ago, I would think there was an error somewhere.” Deposition of J. Frank Stewart, M.D. at p. 13. None of the medical experts testified that the blood gas results were invalid, only that each suspected error upon first viewing the results.
The ALJ, in analyzing whether Goodloe satisfied the section 727.203(a)(3) criteria, noted that the 1984 test, on its face, produced qualifying values for the presumption to be invoked. The ALJ also noted that Peabody challenged the validity of the 1984 blood gas study given that Goodloe was alive and ambulatory rather than receiving oxygen for respiratory failure, as would be indicated by such values. But the ALJ was not convinced:
Even though Drs. Stewart and Howard did, in fact, question the accuracy of the ■1984 blood gas study, they each failed to state unequivocally that the listed P02 of 45 could not have been correct. When deposed, Dr. Stewart ... testified that the P02 reading of 45 indicated to him that there might have been an error in conducting the 1984 study, but that such [a] low reading could have been correct if Claimant continued to be a heavy smoker and did not remain physically active through 1984- Dr. Howard specifically stated that a person would not necessarily have to be on oxygen therapy to have a P02 value of 45. I find that Employer has failed to invalidate the 1984 study. Accordingly, this study producing qualifying values, I find the presumption to be invoked under this subsection of the Regulations.
See Decision and Order Denying Benefits, May 6, 1985 at p. 4 (emphasis in original).
Finding the presumption invoked, the ALJ then analyzed whether Peabody could rebut the presumption pursuant to section 727.203(b). Peabody could rebut the presumption of pneumoconiosis in one of four ways: (1) by showing that Goodloe was doing his usual coal mine work or comparable, gainful work; (2) by showing that Goodloe was capable of doing his usual coal mine work or comparable, gainful work; (3) by showing that the disability Goodloe suffered did not arise in whole or in part out of coal mine employment; or (4) by showing that Goodloe did not have pneumoconiosis. Because Goodloe had not worked in the coal mines for five years, the ALJ found no rebuttal under section (b)(1). However, the ALJ did find rebuttal under section (b)(2), accepting the opinion of Dr. Stewart that Goodloe had no pulmonary impairment sufficient to prevent him from performing his usual coal mine employment. The ALJ expressly accepted the opinion of Dr. Stewart over both Drs. Howard and Peters. Dr. Howard offered his opinion only after being directed to exclude the 1984 blood gas study from his analysis, and thus had not considered all the relevant medical evidence in forming his opinion. Dr. Peters, an osteopath, had made *210several inaccurate statements about fundamental concepts of pulmonary medicine.
The ALJ also found rebuttal under section (b)(3), again giving more weight to the opinion of Dr. Stewart, who indicated that Good-loe’s disability, to the extent he had one, was not related to dust exposure in coal mine employment. Because the weight of the x-ray evidence was negative for pneumoconio-sis, and because Dr. Stewart indicated that Goodloe’s condition was not caused by coal dust exposure, the ALJ further found that Goodloe did not have pneumoconiosis, and thus the presumption was also rebutted under section (b)(4). The ALJ therefore denied Goodloe’s claim for benefits.
Goodloe appealed to the Benefits Review Board (the “BRB”), which issued its Decision and Order on August 31, 1989. The BRB affirmed the ALJ’s invocation of the interim presumption, stating that the ALJ was entitled to credit the 1984 blood gas study because Drs. Howard and Stewart were equivocal concerning its accuracy. But the BRB reversed each of the ALJ’s findings that Peabody had rebutted the presumption, finding that our intervening decision Wetherill v. Director, Office of Workers’ Compensation Programs, 812 F.2d 376 (7th Cir.1987), required reversal of the (b)(2) rebuttal finding. Under Wetherill, the BRB stated, the ALJ
must consider the issue of total disability without regard to cause in concluding whether claimant is able to perform his usual coal mine work or comparable and gainful work. In the instant case, the administrative law judge relied on the opinion of Dr. Stewart that claimant had no pulmonary impairment which was sufficiently severe to prevent him from performing his usual coal mine employment. In light of the Court’s holding in Wetherill,
... Dr. Stewart’s opinion is insufficient to rebut the presumption at subsection (b)(2). We, therefore, vacate the administrative law judge’s finding that rebuttal was established under subsection (b)(2) and remand ... for application of the standard set forth in Wetherill.
Decision and Order of the Benefits Review Board, August 31, 1989, at p. 3 (internal citations omitted). The BRB also reversed the (b)(3) rebuttal finding and ruled that the ALJ had failed to resolve an evidentiary inconsistency relating to the (b)(4) rebuttal finding, reversing that finding as well.
On remand, the ALJ, using the standards set out by the BRB interpreting Wetherill, found that Peabody failed to rebut the presumption under (b)(2) because there was no evidence about whether Goodloe was free of disability generally or was generally capable of performing his usual coal mine (or other comparable, gainful) work. See Decision and Order Awarding Benefits on Remand, February 21, 1992, at p. 2. Rather, Peabody’s evidence focused on whether Goodloe suffered a pulmonary disability that kept him from gainful employment, and the ALJ now believed that Wetherill required the employer to prove the miner suffered no disability, regardless of cause. Similarly, for section (b)(3) rebuttal, the ALJ found that Peabody did not rebut the presumption under the Wetherill standard which required Peabody to “rule out” any causal relationship between Goodloe’s disability and his coal mine employment. Again focusing on Dr. Stewart’s opinion for the reasons stated in his original Order and Decision Denying Benefits, the ALJ found that Dr. Stewart failed to identify the etiology of Goodloe’s pulmonary impairment, and also failed to unequivocally invalidate the 1984 blood gas study.
In addressing the evidentiary inconsistency in the (b)(4) rebuttal, the ALJ agreed with the BRB that Dr. Stewart equivocated on whether the x-ray evidence established pneu-moconiosis, and that Dr. Stewart never identified the etiology of Goodloe’s pulmonary impairment. Because the burden was on Peabody to prove the absence of pneumoco-niosis,' the ALJ found that the presumption was not rebutted under section (b)(4). The ALJ therefore awarded benefits to Goodloe, and allowed Goodloe’s attorney to submit a fee application. Subsequently, the ALJ awarded fees, including enhancements for the contingent nature of the claim and for the delay in payment, to Goodloe’s attorney.
This time, Peabody appealed to the BRB, contesting both the substantive findings of the ALJ and the amount of fees granted to Goodloe’s attorney. In a June 27,1995 Deei*211sion and Order, the BRB noted that Peabody did not challenge the ALJ’s finding that the evidence was insufficient to establish rebuttal under section (b)(2). Rather, Peabody contested the findings under sections (b)(8) and (b)(4), faulting the ALJ for holding Peabody to an erroneously high burden of proof, failing to properly credit the opinion of Dr. Stewart as sufficient for section (b)(3) rebuttal, and failing to consider the testimony of Dr. Howard. The BRB rejected all of these arguments and affirmed the award of benefits. The BRB agreed that the ALJ should not have allowed an increase in the attorneys fees awarded because of the contingent nature of the claims. But the BRB rejected Peabody’s claim that the ALJ could not account for the delay in payment.
On appeal, Peabody claims three main errors in the ALJ’s analysis. First, Peabody faults the ALJ for shifting the burden of proof on the validity of the 1984 blood gas study from Goodloe to Peabody. Rather than requiring the employer to prove the invalidity of the study, Peabody contends that the ALJ should have required the claimant to prove its validity. Second, Peabody argues that the ALJ’s original finding that Goodloe did not suffer a disabling pulmonary impairment was sufficient to establish rebuttal of the interim presumption as a matter of law under section (b)(2). Finally, Peabody contends that the ALJ erred in finding that Peabody -failed to prove that Goodloe did not have pneumoconiosis, and in particular erred by disregarding the opinion of Dr. Howard and not properly weighing the opinion of Dr. Stewart.
Peabody also challenges the award of fees to Goodloe’s attorney, arguing against an increase in the fee awarded due to the contingent nature of the claim or the delay in payment. The Office of Workers’ Compensation Programs (the “OWCP”) of the United States Department of Labor takes the position that a fee award may account for the delay between the time the legal services were performed and the time payment is made, but that the fee may not be enhanced on account of the contingent nature of black lung claims.
Goodloe, in response, disputes that there was any evidence indicating the 1984 blood gas test was invalid. Goodloe further argues that because neither Dr. Stewart nor Dr. Howard considered the 1984 blood gas study in their opinions, their opinions could not support a rebuttal of the interim presumption. Finally, Goodloe contends that the regulations and current case law permit extra compensation for the contingent nature of black lung benefits claims and for the delay between representation and payment.
DISCUSSION
Although the case comes to us on appeal from a decision of the BRB, this Court reviews the decisions of the ALJ, not the BRB. Old Ben Coal Co. v. Battram, 1 F.3d 1273, 1275 (7th Cir.1993). We will not overturn an ALJ’s decision if it is rational, supported by substantial evidence and consistent with the governing law. Id. “Substantial evidence is ‘such relevant evidence as a reasonable mind might accept as adequate to support a conclusion.’” Consolidation Coal Co. v. Office of Workers’ Compensation Programs, 54 F.3d 434, 436 (7th Cir.1995) (citing Freeman United Coal Mining Co. v. Benefits Review Bd., 919 F.2d 451, 452 (7th Cir.1990), modified, Amax Coal Co. v. Beasley, 957 F.2d 324 (7th Cir.1992)). - We may not reweigh the evidence or substitute our own judgment for that of the ALJ, but we must carefully review the record and the ALJ’s findings. Consolidation Coal, 54 F.3d at 436. The ALJ must consider all relevant medical evidence, may not substitute his or her own expertise for that of qualified physicians, and cannot disregard the medical conclusions of a qualified physician unless there is countervailing clinical evidence or a valid legal basis for doing so. Id.
We first address whether the ALJ erred in finding that Goodloe was entitled to the interim presumption.1 The ALJ found *212that Goodloe qualified for the interim presumption on the basis of the 1984 blood gas study, a study that Peabody questioned through its experts. But the ALJ reasoned that because Peabody’s experts “failed to state unequivocally that the listed P02 of 45 could not have been correct,” Peabody “failed to invalidate the 1984 study.” On that basis, the ALJ invoked the presumption.
In requiring Peabody to invalidate the study, however, the ALJ erroneously shifted the burden of proof on the interim presumption from the claimant to the employer. The claimant bears the burden of proving an invocation fact by a preponderance of the evidence. Mullins Coal Co. of Virginia v. Director, Office of Workers’ Compensation Programs, 484 U.S. 135, 158-59 and 162 n. 35, 108 S.Ct. 427, 439-40 and 441 n. 35, 98 L.Ed.2d 450 (1987), reh’g denied, 484 U.S. 1047, 108 S.Ct. 787, 98 L.Ed.2d 872 (1988) (“By requiring miners to show that they suffer from the sort of medical impairment that initially gave rise to congressional concern, and then by requiring employers to shoulder the remainder of the proof burden, the Secretary’s reading of the interim presumption’s invocation burden satisfies both the purpose of the statute and the need for a logical connection between the proven fact and the presumed conclusion.”); Cook v. Director, Office of Workers’ Compensation Programs, 816 F.2d 1182, 1186-87 (7th Cir.1987) (the ALJ need not find that a single positive x-ray establishes the presumption but may consider other, contrary results in determining if claimant has established the presumption). Here, the ALJ acknowledged that the employer’s experts, and even the claimant’s expert, raised questions about the validity of the 1984 blood gas study. By then finding that the employer failed to invalidate that study, the ALJ placed an impermissible burden on the employer. We are therefore remanding so that the ALJ can apply the proper legal standard in determining whether to invoke the interim presumption.
On remand, the ALJ should require Good-loe to prove the validity of the study by a preponderance of the evidence. The ALJ should invoke the presumption if it was more likely than not that the 1984 blood gas study was valid. The ALJ is free, of course, to accept that testimony of Dr. Peters that he called the laboratory where the study was conducted to verify the results, and that the results were verified to his satisfaction. The ALJ may also consider Dr. Peters’ testimony that the physical appearance of the claimant was consistent with a blood gas reading at this level. See Deposition of Henry A. Peters, D.O. at p. 9 (“He said he was short of breath with slight exertion. This was rather obvious, he was breathing rather hard when he got into the office.”); Deposition of David W. Howard, M.D. at p. 5 (“I would expect this person, any person with a P02 of 45 to appear short of breath, especially with exertion ...”). But the ALJ will also have to consider the testimony of all three doctors, detailed above, that they each viewed the results with suspicion.
Because we are remanding on the interim presumption issue, it is premature for us to decide the rebuttal issues or the attorney’s fee question. If the ALJ determines on remand that Goodloe has carried his burden in proving an invocation fact, then the ALJ will have to decide whether Peabody can rebut the interim presumption. In analyzing the rebuttal issue, the ALJ should consider our-decision in Freeman United Coal Mining Co. v. Foster, 30 F.3d 834 (7th Cir.1994), cert. denied, 514 U.S. 1035, 115 S.Ct. 1399, 131 L.Ed.2d 287 (1995), in which we detail our interpretation of Wetherill. In Foster, we revisited whether section (b)(2) permits rebuttal when the miner is not able to perform his usual coal mine work but where the disability is caused by something other than pneumoconiosis. 30 F.3d at 837. We noted that the language of section (b)(2) “looks like a reference to disability pure and simple,” until one examines the cross reference. Id. Section (b)(2) permits rebuttal where “[i]n light of all relevant evidence it is established that the individual is able to do his usual coal mine work or comparable and gainful work *213(see § 410.412(a)(1) of this title).” Section 410.412(a)(1), in turn, defines “total disability” in terms of activities the miner is prevented from engaging in because of his pneu-moconiosis. In Foster, because the miner’s inability to work was due to a back injury and not to pneumoconiosis, we held that the employer established rebuttal under section (b)(2). Thus, we held that the cross-reference in (b)(2) “invites us to consider not only the miner’s ability to do his old job but also whether inability to work is traceable to coal dust.” 30 F.3d at 839. See also Old Ben Coal Co. v. Director, Office of Workers’ Compensation Programs, 62 F.3d 1003, 1008 (7th Cir.1995) (in Foster, “[w]e repudiated our previous Wetherill dicta and held that a miner whose pneumoconiosis was not disabling and who could have worked except for work-related back injury was not totally disabled by pneumoconiosis and was not entitled to benefits.”). Accordingly, we remand with the hope that this case will not enter a third decade of litigation.
REVERSED AND REMANDED.
. As a threshold matter, we must decide whether Peabody waived this argument because it raised the argument only the first time it was before the BRB. We are persuaded by Peabody’s contention that once it raised the issue and the BRB ruled against it, the BRB’s ruling was law of the case. Peabody was not then required to raise the issue again on subsequent appeals before the BRB *212unless a change in the law gave Peabody new arguments that it wished to preserve. Apparently, no new law arose between the first and second appeals to the BRB, and thus the issue was preserved for appeal to this Court.