OPINION OF THE COURT
STAPLETON, Circuit Judge.Three Mile Island Nuclear Station has two units, TMI-1 and TMI-21. On March 28, 1979, an accident severely damaged TMI-2. Prior to the accident, TMI-1 had been shut down for normal refueling and maintenance. On July 2, 1979, the Nuclear Regulatory Commission (“NRC”) issued an *722immediately effective order requiring that TMI-1 remain shut down until the Commission, after a public hearing, determined that there was reasonable assurance that the licensee could restart and operate the unit without endangering the health and safety of the public. 44 Fed.Reg. 40,461 (July 10, 1979). On August 9, 1979, the NRC issued a further order specifying the procedural and substantive format for the hearing. Metropolitan Edison Co. (Three Mile Island Nuclear Station, Unit No. 1), CLI-79-8, 10 NRC 141 (1979).
The petitioners in this review proceeding are the Commonwealth of Pennsylvania (the “Commonwealth”), Three Mile Island Alert, Inc. (“TMIA”), the Union of Concerned Scientists (“UCS”), and Norman and Marjorie Aamodt (the “Aamodts”). These petitioners intervened before the Commission and participated actively and effectively in the extensive hearings conducted pursuant to the August 9, 1979 order. While the issues addressed in these hearings remained under advisement at various levels of the administrative process, several motions were filed asking that the hearing record be reopened. While the record was reopened to receive some additional evidence, the Commission denied motions of these petitioners.
On May 29, 1985, the Commission decided that TMI-1 could be safely restarted under certain stipulated conditions and ordered that the 1979 immediately effective suspension of the TMI-1 operating license be lifted. Petitioners insist that the Commission could not legally take this action without holding an adjudicatory hearing on the issues raised in their motions to reopen the administrative record. Some of the petitioners also contend that the decision reflected in the Commission’s May 29th Order is arbitrary and capricious. We conclude that the May 29th Order is not arbitrary, capricious, or contrary to law and that, accordingly, the petitions for review should be denied.
I. SUMMARY OF THE PROCEEDING
The proceeding that gives rise to these petitions for review is one of the most comprehensive adjudicatory proceedings ever conducted by the NRC. The procedural history of this matter is described in detail by the Commission in its May 29, 1985 order, CLI-85-9, 21 NRC -, and will not be repeated here in full. Nevertheless, a summary of that complex history is a prerequisite to an understanding of the issues raised by petitioners.
The Commission’s August 9, 1979 “Order and Notice of Hearing” (“Order”) called for an adjudicatory hearing to commence within 180 days of publication of the Order and established a Licensing Board, consisting of one lawyer and two scientists,2 to conduct the hearing. CLI-79-8, 44 Fed.Reg. 47821, 10 NRC 141 (1979). The proceedings conducted pursuant to the Order are hereafter referred to as the “Restart Proceedings.” The purpose of the hearing was to determine “whether any further operation [of TMI-1 would] ... be permitted and, if so, under what conditions.” 10 NRC at 142. The Commission directed the Licensing Board to answer two questions:
(1) Whether the “short term actions” recommended by the Director of Nuclear Reactor Regulation ... are necessary and sufficient to provide reasonable assurance that the Three Mile Island Unit 1 facility can be operated without endangering the health and safety of the public, and should be required before resumption of operation should be permitted.
(2) Whether the “long-term actions” recommended by the Director of Nuclear Reactor Regulation ... are necessary and sufficient to provide reasonable assurance that the facility can be operated for the long term without endangering the health and safety of the public, and should be required of the licensee as soon as practicable.
10 NRC at 148.
The Order required the licensee, Metropolitan Edison Company (“Met Ed”), to *723maintain TMI-1 in a “cold shutdown condition” until (1) satisfaction of such short-term actions as the Commission ultimately determined, “after review of the Licensing Board’s decision, to be necessary and sufficient to provide adequate protection of the public health and safety” and (2) reasonable progress toward the long-term actions that the Commission determined to be necessary and sufficient to satisfy the same standard. 10 NRC at 146. The NRC also requested the Director of its staff to determine outside of the adjudicatory proceeding whether the short-term conditions ultimately imposed had been satisfactorily completed. Restart was prohibited until the Director certified to the Commission that all such conditions had been met.
The Commission intended the capability and integrity of the licensee’s operating and managerial personnel (the “management competence” issues) to be important issues in the restart proceedings. Among the “short-term” actions proposed by the Director and subject to adjudication before the Licensing Board were that the licensee, with respect to TMI-1:
Augment the retraining of all Reactor Operators and Senior Reactor Operators assigned to the control room including training in the areas of natural circulation and small break loss of coolant accidents including revised procedures and the TMI-2 accident. All operators will also receive training at the B & W simulator on the TMI-2 accident and the licensee will conduct a 100 percent reexamination of all operators in these areas. NRC will administer complete examinations to all licensed personnel in accordance with 10 CFR 55.20-23.
The licensee shall demonstrate his managerial capability and resources to operate Unit 1 while maintaining Unit 2 in a safe configuration and carrying out planned decontamination and/or restoration activities. Issues to be addressed include the adequacy of groups providing safety review and operational advice, the management and technical capability and training of operations staff, the adequacy of the operational Quality Assurance program and the facility procedures, and the capability of important support organizations such as Health Physics and Plant Maintenance.
10 NRC at 144-45.
Before the hearing commenced, the Commission entered a supplementary order, CLI-80-5, 11 NRC 408 (March 6, 1980), which gave the Licensing Board further guidance regarding the management competence issues:
In determining whether Metropolitan Edison is capable of operating Unit 1 safely, the Board is directed to examine the following broad issues: (1) whether Metropolitan Edison’s management is sufficiently staffed, has sufficient resources and is appropriately organized to operate Unit 1 safely; (2) whether facts revealed by the accident at Three Mile Island Unit 2 present questions concerning management competence which must be resolved before Metropolitan Edison can be found competent to operate Unit 1 safely; and (3) whether Metropolitan Edison is capable of operating Unit 1 safely while simultaneously conducting the clean-up operation at Unit 2.
In the course of examining these broad questions, the Licensing Board should examine the following specific issues:
(1) whether Metropolitan Edison’s command and administrative structure, at both the plant and corporate levels, is appropriately organized to assure safe operation of Unit 1;
(2) whether the operations and technical staff of Unit (sic) is qualified to operate Unit 1 safely (the adequacy of the facility’s maintenance program should be among the matters considered by the Board);
(3) What (sic) are the views of the NRC inspectors regarding the quality of the management of TMI Unit 1 and the corporate management, staffing, organization and resources of Metropolitan Edison;
*724(7) whether Metropolitan Edison has made adequate provision for groups of qualified individuals to provide safety review of and operational advice regarding Unit 1;
(8) what, if any, conclusions regarding Metropolitan Edison’s ability to operate Unit 1 safely can be drawn from a comparison of the number and type of past infractions of NRC regulations attributable to the Three Mile Island Units with industry-wide infraction statistics;
(9) what, if any, conclusions regarding Metropolitan Edison’s ability to operate Unit 1 safely can be drawn from a comparison of the number and type of past Licensee Event Reports (“LER”) and the licensee’s operating experience at the Three Mile Island Units with industry-wide statistics on LER’s (sic) and operating experience;
(10) whether the actions of Metropolitan Edison’s corporate or plant management (or any part or individual member thereof) in connection with the accident at Unit 2 reveal deficiencies in the corporate or plant management that must be corrected before Unit 1 can be operated safely____
CLI-80-5, 11 NRC 408, 409 (1980).
The August 9, 1979 Order required the Licensing Board to render decisions on these issues and to certify the record to the Commission for final decision.3 The Order further provided for an unusual two-track review of the Licensing Board’s decision.4 In addition to appellate review of Licensing Board decisions by the Commission, the Order created an “effectiveness review” to determine whether to lift the immediate effectiveness of the shutdown orders and to permit operation of the plant prior to completion of the appellate review.
The Licensing Board thereafter conducted formal hearings that in their detail and complexity far exceeded the Commission’s original expectations. Whereas the Commission anticipated that the hearings would take 60 days and that the Licensing Board would render its decision eleven months after the August, 1979 Order, the Licensing Board ultimately held 155 days of hearings and, six years after the TMI-1 shutdown orders, has yet to issue the final installment of its decision.
The comprehensiveness of the NRC proceeding can be measured in part by the enormity of its record. The administrative record currently consists of more than 100,-000 pages, including approximately 33,000 transcribed pages of testimony and argument and thousands of additional pages of exhibits and written testimony. The Licensing and Appeal Boards and the Commission have published opinions filling 1,500 pages in the NRC’s official reports. The Commission itself has heard oral presentations from the parties on five different occasions, has held a meeting in Harrisburg, Pennsylvania, to hear from members of the public, and has issued more than 26 substantive orders.
At the hearings themselves, all parties had the opportunity to cross-examine witnesses and offer rebuttal evidence. The Licensing Board conducted an extensive inquiry regarding the management competence issues. The licensee produced its management and technical personnel at the hearing, and petitioners and other intervenors, as well as the Board, examined them at length. LBP-81-32, 14 NRC 381, 401 (August 27, 1981). Moreover, at a *725supplementary hearing before a Special Master on allegations that TMI-1 plant operators had cheated on tests, the operators testified under a sequestration order and were thoroughly examined. LBP-82-56, 16 NRC 281, 291 (July 27, 1982).
The Licensing Board has issued four partial initial decisions, each of which- — after imposing certain conditions on restart— concludes that TMI-1 may be restarted. In its first decision, issued August 27, 1981, the Board addressed several management-related issues, including licensee’s management structure, the adequacy of licensee’s training program, licensee’s response to the accident at TMI-2, and licensee’s capability and resources. LBP-81-32, 14 NRC 381 (1981). The Board concluded that the Director’s recommendations governing management competence were necessary and sufficient to ensure public safety, but retained jurisdiction to consider the effect on its decision of an ongoing investigation of allegations of operator cheating on licensing exams.5 The Board noted that because it lacked sufficient information, it had not considered in its decision the subject matter of a United States Department of Justice criminal investigation of alleged falsification by TMI-2 operators of safety-related leak rate tests. On October 2, 1981, the Board reopened the record on the implications of the alleged operator cheating and appointed a Special Master to hear the relevant evidence.
The Licensing Board issued its second initial decision on December 14, 1981, covering hardware and design issues, the separation of Units 1 and 2,6 and emergency planning. LBP-81-59, 14 NRC 1211 (1981). Again, the Board found that the licensee satisfied NRC short-term requirements, subject to correction of several deficiencies in design, procedures and planning. These corrections, “in the form of Licensee commitments, NRC Staff requirements and Board-imposed conditions,” provided reasonable assurance that TMI-1 could be operated without endangering the health and safety of the public. 14 NRC at 1711. The Board also determined that licensee had made reasonable progress toward the Director’s recommended long-term actions.
Before the Licensing Board issued its next partial initial decision, licensee undertook a reorganization that resulted in Met Ed being replaced by GPU Nuclear Corporation (“GPUN”) as the operator of TMI-1. GPUN was a newly-created subsidiary of General Public Utilities Corporation (“GPU”) whose sole responsibility is management of GPU’s three nuclear power plants.7
The Board’s third partial initial decision, issued on July 27, 1982, addressed the allegations of cheating in response to which it had reopened the record the previous October. LBP-82-56, 16 NRC 281 (1982). This decision reviewed the Special Master’s report, which had been submitted on April 28, 1982, LBP-82-34B, 15 NRC 918 (1982), imposed five new conditions on the restart of TMI-1, and concluded that the favorable determinations of the first and second partial initial decisions remained in effect. 16 NRC at 385.
After the Licensing Board rendered its third partial initial decision, litigation between GPU and Babcock & Wilcox Co., the manufacturer of the TMI-2 reactor, produced information that petitioners believe highly relevant to the management competency issues. That information, according to petitioners, indicates that the licensee made false statements to the NRC in its response to a Commission Notice of Violation for misconduct related to the TMI-2 accident. On April 18, 1983, the NRC staff *726announced it was “revalidating” its earlier position on management competence.8
Soon thereafter, GPUN committed to make three significant organizational changes. On June 10, 1983, it announced it would realign personnel to minimize involvement at TMI-1 of employees who had preaccident involvement at TMI-2, add full-time, on-shift operational quality assurance coverage by degreed engineers, and realign functions within the office of the president of GPUN.
On August 31, 1983, the Appeal Board decided to reopen the record for further hearings on the impact of the leak rate falsification charges on management competence. ALAB-738, 18 NRC 177 (1983). The Commission, in an October 7, 1983 unpublished order, took review of the Appeal Board’s determination and stayed any hearings into the matter.
The Commission heard oral presentations from all parties on November 28 and December 5 regarding the reorganization of the licensee. At the November hearing, GPU announced further changes in its organization. Besides verifying that the June 10th plan had been implemented, the company announced that three outside directors had been added to the GPUN Board of Directors, and that they would comprise a separately staffed and funded Nuclear Safety and Compliance Committee. GPU also announced that the President of GPUN, Robert Arnold, had resigned. On February 6, 1984, GPU announced further personnel changes, including one affecting Herman Dieckamp, who, while continuing to serve as GPU’s President, would no longer serve as Chairman and Chief Executive Officer of GPUN.9
The cumulative effect of personnel changes at TMI-1 since the TMI-2 accident was to significantly alter the GPUN workforce. Of the twelve senior GPUN officers, eight joined the GPU system after the TMI-2 accident, and three of the remaining four had no connection with Met Ed. Of 435 key personnel at GPUN, 235 joined GPUN after the accident, and 100 preaccident employees had never worked with Met Ed. The number of full-time employees who are to work at TMI-1 is almost triple the number employed when Met Ed operated it (915 to 315). Finally, GPUN employs nearly eight times as many employees responsible for training as did Met Ed (55 to 7). CLI-85-09, 21 NRC -, - n. 39 (1985).
On February 28, 1984, Met Ed pled guilty to criminal falsification of leak rate data. A Statement of Facts submitted to the Court by the United States Attorney in connection with the sentencing specified that although “the evidence would establish that a number of employees of the Metropolitan Edison Company engaged in the criminal activities charged in the indictment,” “the evidence presented to the grand jury and developed by the United States Attorney does not indicate that any” of the directors or officers of GPUN from its inception in 1982 “participated in, directed, condoned, or was aware of the acts or omissions that are the subject of the indictment.”
On May 24, 1984, the Appeal Board reversed the Licensing Board’s third partial initial decision and reopened the record on three issues. ALAB-772, 19 NRC 1193 (1984) . These were the effect of the cheating incidents on training, the truthfulness of a mailgram sent by GPU President Herman Dieckamp to Congressman Morris Udall shortly after the accident, and the possibility of improper leak rate practices at TMI-1. 19 NRC at 1279-80.
The Commission on February 25, 1985, reversed the Appeal Board’s decision to reopen the record: CLI-85-2, 21 NRC 282 (1985) . After considering the numerous contentions raised by the parties, the Commission found that none of those contentions warranted further hearings in the *727restart proceeding. The Commission in particular addressed the petitioners’ claim that leak rate falsifications at TMI-2 raised sufficient doubts regarding management competence as to require reopening of the record. It concluded that personnel changes by the licensee mooted the immediate significance of the TMI-2 leak rate manipulation and, to provide added assurance, imposed the following conditions on the licensee:
(1) No pre-accident TMI-2 operator, shift supervisor, shift foreman, or any other individual both in the operating crew and on shift for training as a licensed operator at TMI-2 prior to the accident shall be employed at TMI-1 in a responsible management or operational position without specific Commission approval.
“Operational position” as used here includes any position involving actual operation of the plant, the direction or supervision of operators, or independent oversight of operations.
This condition shall also apply to the pre-accident Vice President, Generation, TMI-2 Station Manager, TMI-2 Supervisor of Technical Support (from January 1977 to November 1978), TMI-2 Supervisor of Technical Support (from December 1978 to the accident), and TMI-2 Supervisor of Operations. This condition shall not apply to Michael Ross, and Brian Mehler may continue in his present position consistent with this condition.
(2) Licensee, in the absence of Commission authorization to the contrary, is to retain its expanded Board of Directors and its Nuclear Safety and Compliance Committee.
21 NRC at 341-2.
The Commission’s February 25th order also initiated a new proceeding to consider disciplining employees, other than those explicitly cleared by the United States Attorney in his Statement of Facts, possibly involved in leak rate falsification at TMI-2.10 21 NRC at 287. Finally, the Commission directed the Licensing Board to issue decisions on the training and Dieckamp mailgram issues, on which the Board already had completed evidentiary hearings.
The Licensing Board issued its fourth partial initial decision on May 3, 1985. LBP-85-15, 21 NRC-(1985). This decision reviewed the adequacy of licensee’s training programs in light of the cheating incidents. The Board, after requiring the licensee to implement a plan for on-the-job performance evaluation, once again found in favor of restart.
On May 9, 1985, the Director of the NRC certified that the licensee had satisfied' all of the 155 conditions that had been imposed during the Restart Proceedings, and on May 29, 1985, the Commission, after first reiterating its conclusion that personnel and procedural changes at TMI-1 had rendered the allegations raised by petitioners moot for the purposes of determining when and under what conditions restart could safely occur, lifted the immediate effectiveness of the orders requiring the shutdown of TMI-1. CLI-85-09, 21 NRC-(1985).
II. THE SCOPE OF REVIEW
Pursuant to 5 U.S.C. § 706(2)(A), the standard of review of the Commission’s order of May 29, 1985, is deferential; that order may not be overturned unless it is found to be “arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.”
With respect to petitioners’ procedural arguments, this Court has noted that the NRC regulatory scheme is “virtually unique in the degree to which broad responsibility is reposed in the administering agency, free of close prescription in its charter as to how it shall proceed in achieving the statutory objectives.” Westinghouse Electric Corp. v. Nuclear Regulatory Commission, 598 F.2d 759, 771 (3d *728Cir.1979), quoting Siegel v. Atomic Energy Commission, 400 F.2d 778, 783 (D.C.Cir.1968). See also San Luis Obispo Mothers for Peace v. Nuclear Regulatory Commission, 751 F.2d 1287, 1294 (D.C.Cir.1984); Carstens v. Nuclear Regulatory Commission, 742 F.2d 1546, 1551 (D.C.Cir.1984). A narrow standard is particularly appropriate in reviewing petitioners’ central challenge, an attack on the Commission’s refusal to reopen the record of the Restart Proceedings to hear further evidence on certain issues. As the United States Court of Appeals for the District of Columbia recently held, “[w]here as here the agency has taken final action on a matter that is peculiarly within its realm of expertise, we will not require the agency to reopen its proceedings except upon a clear showing of abuse of discretion or of extraordinary circumstances.” San Luis Obispo, supra, at 1317-18, quoting Mobil Oil Corp. v. Interstate Commerce Commission, 685 F.2d 624, 632 (D.C.Cir.1982).
With regard to the claim raised by petitioners Aamodts that the TMI-2 accident caused greater amounts of radiation to be released than the NRC believes, deference must be accorded to the Commission’s expertise. “When examining this kind of scientific determination, as opposed to simple findings of fact, a reviewing court must generally be at its most deferential.” Baltimore Gas & Electric Co. v. Natural Resources Defense Council, Inc., 462 U.S. 87, 103, 103 S.Ct. 2246, 2256, 76 L.Ed.2d 437 (1983).
Our “limited, albeit important, task” is to review “agency action to determine whether the agency conformed with controlling statutes.” Id. at 97, 103 S.Ct. at 2252. In undertaking this task, we must be mindful both of the Congressional policy choice favoring the development of nuclear power and of the Congressional mandate that the Commission be charged with the responsibility of protecting the public health and safety while overseeing the activities of the nuclear power industry. Our words in Westinghouse Electric Corp. v. Nuclear Regulatory Commission are equally appropriate here:
Congress has delegated authority in the delicate area of nuclear energy to a number of agencies, among them the NRC. The NRC is charged with the responsibility of protecting the common defense and security as well as the public health and safety, while overseeing the licensing of nuclear facilities. Some of the decisions it makes to further its statutory mandate may be unpopular in the nuclear industry, among environmentalists, or with other groups of citizens. But Congress has decreed that the agency be independent from outside control, and it would subvert this design were we to invalidate the challenged NRC action when it appears to be consonant with statutory dictates and not an unreasonable exercise of its discretion.
Westinghouse Electric Corp., supra, 598 F.2d at 779.
III. CONSISTENCY OF THE PROCEEDINGS WITH § 189(a) OF THE ATOMIC ENERGY ACT
The Restart Proceedings resulted in 155 conditions being imposed on the licensee to ensure that TMI-1 could be operated consistent with public health and safety. Petitioners argue that the procedures followed by the NRC in this case, which led to imposition of these conditions, violate Section 189(a) of the Atomic Energy Act, 42 U.S.C. § 2239(a) (1976) (amended by Pub.L. No. 97-415, § 12, 96 Stat. 2067, 2073 (1983)). Section 189(a) provides:
(1) In any proceeding under this chapter, for the granting, suspending, revoking, or amending of any license or construction permit, or application to transfer control, and in any proceeding for the issuance or modification of rules and regulations dealing with the activities of licensees ... the Commission shall grant a hearing upon the request of any person whose interest may be affected by the proceeding, and shall admit any such person as a party to such proceeding____
The petitioners insist that the Restart Proceeding is a proceeding for the amending of a license within the meaning of Section *729189(a) and that they are persons whose interest stands to be affected by that proceeding. The NRC maintains that the Restart Proceeding is an enforcement proceeding rather than a proceeding for amending a license, that petitioners are not persons whose interests could be affected by the addition of more restrictive safety conditions, and that, in any event, the Commission has voluntarily accorded the petitioners all the procedural protection to which they would have been entitled if Section 189(a) were applicable.
Under 10 CFR § 2.200-.206 (1985), the Commission is authorized to institute enforcement proceedings “to impose requirements by order, or to modify, suspend, or revoke a license, or to take other action as may be proper.” 10 CFR § 2.200 (1985).11 When such a proceeding is instituted, the licensee ordinarily continues to operate during the pendency of the proceeding pursuant to its outstanding authority. In some circumstances, however, the Commission may determine that the “public health, safety, or interest” requires “that the proposed action be temporarily effective pending further order” of the Commission. 10 CFR § 2.202(f) (1985). The effect of these regulations is to provide the Commission with the authority, where issues of public health and safety are involved, to determine whether or not a licensee will be permitted to continue to operate during an enforcement proceeding and, if so, under what conditions. In this case, the Commission decided to immediately shut down TMI-1 until public hearings on conditions of operation were held.
As noted above, the August 9, 1979 Order spelled out a dual review procedure. It designated the appellate review for resolution of the ultimate issues in the enforcement proceeding and the immediate effectiveness review for resolution of the issues of whether, when and under what conditions the “immediate effectiveness” of the July 2, 1979 shutdown order would be lifted. Although the Commission’s May 29, 1985 resolution of the “immediate effectiveness” issues came later in the proceedings than originally contemplated, it did not terminate those proceedings. The Commission’s May 29, 1985 Order is thus a determination relating solely to whether the licensee will be able to utilize its existing authority during the pendency of an ongoing Commission proceeding.
This procedural context is important. While amendments to the TMI-1 operating license may, and probably will, be imposed before the end of the current proceeding, the May 29, 1985 Order does not purport to impose such amendments. The sole effect of that Order is to lift the 1979 shutdown orders; the licensee has no greater operating authority by virtue of the May 29, 1985 Order than it had on July 1, 1979. It is true that the Commission, as a condition of lifting its shutdown order, has imposed numerous restrictions on the way in which the licensee may exercise its existing authority, but petitioners have pointed us to nothing in this record which indicates that the Commission has purported to effect amendments to the license or that license amendments are necessary to permit the licensee to operate in accordance with the restrictions which have been imposed.12 The imposition of conditions on the lifting of the stay and thus on the licensee’s interim operations does not establish that the Commission has purported to amend the outstanding license since the Commission’s regulations contemplate that such restrictions may be imposed during the course of a proceeding to determine whether or not license amendments or other actions are appropriate.13
*730Because the May 29, 1985 Order does not purport to effect a license amendment, the proceeding before us does not present the primary procedural issue which petitioners seek to litigate- — whether the Commission must comply with Section 189(a) of the Atomic Energy Act before amending a license in an enforcement proceeding. Accordingly, we express no opinion on that issue. We hold only that Section 189(a) is not implicated when the Commission enters an order lifting a suspension so that a licensee may operate under existing authority during the course of an enforcement proceeding.14
IV. CONSISTENCY OF THE PROCEEDINGS WITH THE ORDER OF AUGUST 9, 1979.
Petitioners contend that the May 29, 1985 decision lifting the immediate effectiveness of the shutdown orders is inconsistent with the procedural rules established by the Commission in its August 9, 1979 Order. They argue that the 1979 Order precluded the Commission from basing the restart decision on evidence other than that contained in the adjudicatory record developed by the Licensing Board, and that the Commission violated this stipulation by relying on extra-record material to support its May 29th decision. UCS also contends that the 1979 Order precluded the Commission from authorizing restart in the absence of prior favorable determinations by the Licensing and Appeal Boards, and that the Appeal Board’s remand on the three management integrity issues indicates that such favorable determinations are lacking. We discuss each of these arguments in turn.
Extra-Record. Material
We agree with the petitioners that the August 9, 1979 order contemplated immediate effectiveness review by the Commission based on the adjudicatory record developed by the Licensing Board. We assume, without deciding, that this irrevocably committed the Commission to make its effectiveness decision on the basis of that record. Moreover, while the May 29th decision is based upon that adjudicatory record, we agree with petitioners that it rests in part on extra-record facts. However, given that these extra-record facts are undisputed and that petitioners were afforded a fair opportunity to offer both argument and evidence on all of the relevant issues in the Restart Proceeding, we conclude that the Commission’s reliance on extra-record materials in its May 29th decision does not require further Commission proceedings.
While the Commission declares in its decision that it is relying solely on the “formal adjudication record," CLI-85-09, 21 NRC at -, this is not the case. In addition to evidence developed before the Licensing Board, the Commission gave significant weight to facts concerning the reorganization of the licensee which were not developed before that Board. Petitioners do not assert that these extra-record facts are untrue. They argue, however, that *731because of the continuing presence at GPU of three individuals who were employed by it prior to the TMI-2 accident, the personnel changes are not as significant as the Commission suggests. The three employees to whom they refer are William Kuhns, Chairman and Chief Executive Officer of GPU, Herman Dieckamp, President of GPU, and Michael Ross, TMI-1 Manager of Plant Operations. Petitioners suggest that the Commission cannot in fairness rely upon these personnel changes without affording them an additional public hearing at which they will have the opportunity to introduce additional evidence relevant to the integrity of Messrs. Kuhns, Dieckamp, and Ross.
We find this suggestion unpersuasive in light of the fact that the effect of the undisputed facts relied on by the Commission, and the complementary condition insulating TMI-1 operations from virtually all personnel connected with TMI-2 prior to the accident, was to narrow the relevant issues remaining before the Commission and not to raise new ones. Put another way, during the 155 days of hearings, petitioners were well aware that they faced the prospect of having Ross, Kuhns, and Dieckamp associated with the licensee upon restart. They had ample opportunity to cross-examine these men and offer evidence concerning their past. Mr. Ross, for example, appeared before the Licensing Board five times over several days and testified on a broad variety of issues. LBP-81-32, 14 N.R.C. 381, 439 (1981).
While petitioners now feel they have more cross-examination and rebuttal material than they had at the time of the hearing, that is not an uncommon occurrence in adjudicative proceedings, and whether that ammunition may be used is a question to be decided, as it was, in a proceeding on a motion to reopen the record. We conclude that the Commission, having afforded petitioners the opportunity to offer evidence regarding Kuhns, Ross and Dieckamp, could rely upon the undisputed personnel changes without reopening the record on the integrity of these men.
The Need for Prior Favorable Board Determinations
UCS argues that the Commission could not authorize restart of TMI-1 until the Licensing Board and the Appeal Board had first reached favorable determinations on the adequacy of the proposed short-term actions, including those related to management integrity. The Licensing Board’s positive assessment of management competence, UCS points out, was rendered without hearings on the leak rate falsifications at TMI-2 and was made specifically “subject to this matter.” LBP-81-32, 14 NRC 381, 557 (1981). Moreover, UCS stresses that the Appeal Board, when deciding to reopen the record to pursue the leak rate test issue, determined that it should not assess management competence without that evidence. ALAB-738, 18 NRC 177, 190 (1983).
We may assume for present purposes that the August 9, 1979 Order conditioned restart on a prior favorable determination by the Licensing Board, for UCS’s argument fails on its own terms. The record indicates that the Licensing Board has in fact issued a favorable determination on management integrity. Although the Appeal Board believed that further hearings were desirable, the Appeal Board’s decision to reopen the record was reversed by the Commission. We find nothing in the August 9 Order to indicate that the Commission surrendered its right to make the ultimate determination on whether and when to reopen the record. Because the Commission’s reversal of the Appeal Board leaves standing the initial favorable Licensing Board determination, even under the UCS analysis the Commission could lawfully authorize restart.15
*732V. REFUSAL TO REOPEN THE RECORD
As we have noted, the parties extensively litigated the management competence issues before the Licensing Board closed the hearing record and issued its August 27, 1981 decision favorable to the licensee. LBP-81-32, 14 NRC 381 (1981). After the record closed, however, there were several potentially significant developments. The most important of these was Met Ed’s guilty plea to federal criminal charges of leak rate falsification at TMI-2. Petitioners moved to reopen the record on management competence, and the Appeal Board granted their motions on four issues. ALAB-738, 18 NRC 177 (1983); ALAB772, 19 NRC 1193 (1984). After receiving briefs from the parties, the Commission decided, on February 25, 1985, that the hearing record need not be reopened prior to restart: CLI-85-2, 21 NRC 282 (1985). Petitioners insist that the Commission’s refusal to hold hearings on post-record developments renders the May 29, 1985 decision arbitrary and capricious.
In refusing to reopen the record, the Commission engaged in its traditional tripartite inquiry for determining if new evidence is of sufficient importance to outweigh the general policy against permitting reopenings. The Commission asked:
(1) Is the motion timely; (2) does it address significant safety ... issues; and (3) might a different result have been reached had the newly proffered material been considered initially.
21 NRC at 285, n. 3. This test has been judicially sanctioned, San Luis Obispo Mothers for Peace v. N.R.C., 751 F.2d 1287, 1316-18 (D.C.Cir.1984).
At the outset, we reject petitioners’ contention that the Commission cannot rely on extra-record material in assessing the significance of evidence submitted in support of a motion to reopen the record. “Newly proffered material” is by definition extra-record. Because it is also “new” by definition, in the vast majority of cases, limiting the Commission’s consideration of a motion to reopen to data already in the record would require that the moving party’s allegations be accepted at face value. We think such a rule arbitrary and unworkable. If the Commission has, or can obtain through investigation, information bearing on the subject matter of a motion to reopen, we conclude that it should be free to use that information in deciding that motion. Here, for example, the Commission, confronted with a motion that heavily stresses Met Ed’s guilty plea, should not be required, as petitioners would suggest, to ignore the fact that the same investigation that led to that plea also produced information indicating GPUN’s management personnel had been unaware of the criminal activity.
Our resolution of this issue is of considerable importance in this case, for although the several issues raised by petitioners are extra-record in the sense that they do not appear in the Licensing Board’s formal adjudicatory record, they have been carefully examined by NRC staff investigators. In particular, the Commission heavily relied on NUREG-0680, Supp. No. 5, a staff report that reviews the investigations conducted by the NRC Office of Investigation (“OI”) and reaches results favorable to current management on the issues raised by petitioners, including leak rate falsification at TMI-1 and TMI-2.
We now proceed to consider whether the Commission abused its discretion in refusing to reopen the record for each issue raised by petitioners. Although we discuss each issue individually, we have considered the cumulative effect of all of the information available to the Commission, as did it, and our conclusion that the Commission did not abuse its discretion has been reached in that context.
TMI-2 Leak Rate Falsification
By far the most important of the reopening issues is whether the record on *733management competence should be reopened for hearings on the pre-accident falsification of leak rate data by TMI-2 operators in violation of NRC regulations.16 Harold Hartman, a TMI-2 control room operator prior to the accident, alleged that plant operators, with the knowledge of shift supervisors, had falsified tests that measured whether primary system leakage surpassed technical specifications in the TMI-2 license. In April, 1980 the NRC referred the matter to the Department of Justice for criminal investigation. This investigation culminated in the February 28, 1984 guilty plea of Met Ed.
Although we recognize the importance of the issue, we are unable to say that the Commission acted arbitrarily or capriciously in refusing to reopen the record on the leak rate tests. It is undisputable that leak rate falsification had been a serious problem at TMI-2 prior to the accident and further investigation of that problem was clearly called for. The issue was whether to pursue that potentially lengthy process in a new proceeding or whether to reopen the proceeding that had already resulted in TMI-1 being closed down for five and a half years. If the leak rate issue was safety significant to the current operation of TMI-1, the additional delay and attendant expense would, of course, be justified. If that issue was not safety significant to the current operation of TMI-1, further delay and expense would be difficult to justify. In this context, the Commission framed the issues for decision as follows:
To determine whether the Hartman allegations still raise a significant safety issue, the Commission must first consider whether the personnel likely respon-. sible for the falsifications under Met. Ed. are now in responsible management positions at GPU Nuclear or directly associated with the operation of TMI-1. If the personnel are still the same, then there is merit to the argument that there has been a change only in name, not in substance, and the integrity concerns raised by the Hartman allegations remain significant. However, if the persons likely responsible for or involved in the TMI-2 leak rate falsifications are not assigned to responsible management or operational positions at TMI-1, then the Hartman allegations no longer raise concerns about the integrity of those who will operate TMI-1. In that event, however, the Commission' further should consider whether the new personnel, organizational structure, and procedures provide reasonable assurance that similar procedural violations will not recur.
21 NRC at 297.
The Commission concluded that no one assigned to a responsible management or operational position at TMI-1 was responsible in any way for leak rate falsification at TMI-2 and that new personnel, organizational structure and procedures provided reasonable assurance that similar violations would not recur.
The Commission, as a result of the lengthy proceedings before it and the investigation of its staff, was intimately familiar with the events leading up to the accident at TMI-2, with the operating and managerial personnel of the licensee prior to the accident and with the operating and managerial personnel of the licensee prepared to restart TMI-1. The Commission was also intimately familiar with the extensive changes in everything from hardware to training which had been effected at TMI-1 after the accident and with the 153 conditions that had already been imposed. In addition, the Commission was aware *734that the investigation which had uncovered the leak rate falsification problem had indicated that current GPUN management personnel were unaware of that activity and that current operating and supervisory personnel at TMI-1, with one possible exception, had had no connection with pre-accident TMI-2. Finally, the Commission realized that it could assure a continuing absence of taint in the operating personnel at TMI-1 simply by imposing an additional condition, and it did so. Under these circumstances, we conclude that the Commission could reasonably find, as it did, that altered conditions rendered the fact of TMI-2 leak rate falsifications no longer safety significant to TMI-l’s operation.
Petitioners’ argument that the Commission acted arbitrarily or capriciously necessarily focuses on the three individuals who will have some responsibility with respect to TMI-1 who also had some pre-accident involvement with TMI-2: Michael Ross, William Kuhns, and Herman Dieckamp.17 The Commission recognized this fact and set forth the reasons it believed the continued presence of these men did not detract from the assurance provided by the changed conditions at TMI-1.
Michael Ross, as current Manager of Operations at TMI-1, is the only one of the three to hold a safety significant “operational position” at the plant. 21 NRC at 298. Because of the position he would hold, as earlier noted, Ross was one of the focal points of the hearings and extensive findings were made about his competence, including his integrity in the context of alleged operator cheating on tests. Nevertheless, in response to petitioners’ motion to reopen, the 01, at the Commission’s direction, conducted an investigation to determine whether Ross had in any way participated in or condoned leak rate manipulation. As the Commission noted, the “evidence developed by 01 showed that Ross’ role at TMI-2 was minimal, that during the period falsifications took place he was present at TMI-2 only the minimum time necessary to maintain his Unit 2 license, and that he was not involved in the falsifications.” 21 NRC at 298-9. The Commission, accordingly, determined that Ross posed no risk to the safe operation of TMI-1. The only additional information which petitioners advance in support of their position is that Ross has a reputation for being a “stickler for details.” Petitioners infer from this that, despite his minimal contact with TMI-2, Ross must have known that leak rate manipulation was going on there. While this is perhaps a permissible inference, it is not a necessary one, particularly in light of the other information about Ross available to the Commission. Accordingly, we are unable to find that the Commission acted arbitrarily in concluding that the continued presence of Ross at TMI-1 did not require the Commission to reopen the record on TMI-2 leak rate falsifications.
The Commission considered as well the safety significance of the continued presence of Messrs. Kuhns and Dieckamp at GPU. Kuhns has been Chairman of GPU during all relevant periods. Dieckamp remains a member of the Board of GPUN, as well as President, Chief Operating Officer, and a Director of GPU. The Commission found it highly unlikely that corporate officers of their rank would have been aware of the details of normal plant operation such as leak rate testing, and also concluded that Kuhns and Dieckamp were not responsible for the attitude at TMI-2 that permitted the falsifications to occur. 21 NRC at 301. The Commission also relied on the exculpatory statement offered by the United States Attorney in connection with the sentencing hearing of Met Ed:
[T]he evidence presented to the Grand Jury and developed by the United States Attorney does not indicate that any of the following persons participated in, directed, condoned or was aware of the acts or omissions that are the subject of *735this indictment. And they are Williams G. Kuhns, Herman M. Dieckamp____
Transcript of Proceedings, Change of Plea and Sentencing, United States v. Metropolitan Edison Co., Crim. No. 83-00188, at 16 (M.D.Pa. Feb. 28-29, 1984). We believe the Commission could reasonably rely on this statement in determining that the presence of Kuhns and Dieckamp at TMI-2 did not make the leak rate falsifications at TMI-2 safety significant to TMI-1 operations.
In summary, the Commission made a policy decision between holding hearings on the TMI-2 pre-accident, leak rate manipulation problem in the TMI-1 restart proceeding or in a separate proceeding. It chose the latter course because it concluded that changed personnel and altered procedures at TMI-1 rendered the pre-accident experience at TMI-2 irrelevant to an assessment of the safety of TMI-1 operations. In making this choice, the Commission analyzed the relevant issues relying on information from reliable sources, reasoned to a logical conclusion, and articulated the reasons for its decision. As a reviewing court, we can ask nothing more of the Commission.
TMI-1 Leak Rate Test Practices
Petitioners allege that leak rate falsifications also occurred at TMI-1, and that the Commission erred in refusing to reopen the record for hearings on this issue. These allegations arise from the results of an NRC staff review of TMI-1 leak rate tests for the period April 1, 1978 to March 31, 1979. This review, which found 38 of 645 tests had been improperly conducted, led to an investigation by the 01. After the 01 reported, the Appeal Board decided to reopen the record for hearings on TMI-1 leak rate data. ALAB-772,19 NRC 1193 (1984).
The Commission, in reversing the Appeal Board, again relied on the 01 investigation. This investigation included sworn interviews of all pre-accident and current TMI-1 employees who conducted the leak rate tests during the period in question. 21 NRC at 311. The Commission distinguished the leak rate falsifications at TMI-2 from the leak rate “irregularities” at TMI-1 because only three of the 38 questionable tests, if properly conducted, would have indicated excessive leakage. There was therefore no need to falsify test results. Id. Based upon the 01 investigation, the Commission concluded that the intervenors had an unrealistic perspective on the significance of the available data:
The 01 investigation revealed that leak rate testing was considered a ministerial monitoring duty at TMI-1 prior to the accident. It was not difficult to obtain an acceptable leak rate test, and most operators relied on other plant parameters to determine actual leakage. Hence little significance was given to leak rate testing practices. Under these circumstances, it is not surprising that some irregularities in the data may appear. There is no evidence of falsification beyond speculative inferences that could be drawn solely from the circumstance of a few irregularities in the data and some superficial similarity between leak rate practices at TMI-2 and TMI-1. The Commission finds the circumstantial evidence indicating there might have been a few improper leak rate tests conducted in 1978-79 does not raise a significant safety concern that might have changed the Licensing Board’s decision.
Id. at 312 (footnotes omitted).
The Commission also specifically addressed and rejected the possibility that Michael Ross knew of the leak rate irregularities. Id. at 313. The Commission concluded that, given the thoroughness of the 01 investigation, there was no reason to believe that further hearings would produce significant new information on leak rate test irregularities at TMI-1. Id. at 311.
We agree with UCS that the fact that TMI-1 operators did not need to falsify test results does not exculpate them if they in fact did so on occasion. But we understand the Commission to mean that the lack of motivation to falsify is some evidence that the leak rate irregularities un*736covered by the staff review do not reflect a safety-significant pattern of falsification.
While the decision on whether to reopen the record for a hearing on TMI-1 leak rate testing was one about which reasonable minds could differ, as did the Commission and the Appeal Board, this does not make the Commission’s decision arbitrary or capricious. Once again, it responsibly addressed the relevant issues and drew permissible conclusions from all of the data available to it.
Response to the 1979 Notice of Violation (NOV) and Changes in the Keaten Report
The Commission also refused to reopen the record for rehearings on the accuracy of GPU’s response to the NRC’s October, 1979 NOV, which proposed fining the licensee for procedural violations related to the TMI-2 accident. After the accident, GPU created an internal task force, headed by R.W. Keaten, to investigate the accident. Upper management personnel read and revised drafts of the Keaten Report at the same time that licensee submitted its NOV response. After the Keaten Report was completed, the 01 interviewed more than 25 people to determine whether GPU management improperly revised the report. In addition, review of the Keaten Report raised the question of whether licensee knowingly made inaccurate statements in its NOV response. The NRC staff reviewed the 01 investigation and concluded that the NOV response contained statements “neither accurate nor complete and ... contrary to other information in the possession of the licensee.” NUREG-0680, Supp. No. 5 at 8-19. Finally, the staff placed responsibility for the inaccuracies on two officers, one E. Wallace, and the then President of licensee, R.C. Arnold, both of whom no longer are associated with TMI-1, and on Dieckamp, “who reviewed the response before it was submitted and chose ‘not to intervene.’ ” 21 NRC at 330, quoting NUREG-0680, Supp. No. 5 at 8-21.
The Commission, conceding that some changes to the Keaten Report seemed clearly designed to improve licensee’s image, nevertheless did not believe that the changes raised “integrity concerns” that warranted further hearings in the absence of “a showing that false information was used negligently or intentionally.” 21 NRC at 333. Here, the only false factual information the Commission found was that which was also included in the NOV response. It concerned the role played by a relief valve in the TMI-2 accident. The source of the false information was Wallace. Since it was not unreasonable for other officers to rely on Wallace under the circumstances, the Commission found the issue no longer safety significant. Furthermore, the Commission decided that because the Licensing Board had not relied on GPU testimony in any event, further information on GPU’s view of the accident would not be likely to change the Board’s initial favorable determination.
The Commission also decided that questions concerning GPU’s response to the NOV did not justify reopening the record. The Commission reiterated the fact that the two persons most culpable for the inaccuracies in the response — Wallace and Arnold' — were no longer associated with TMI-1. Dieckamp thus formed the only link between the falsehood in the NOV response and the issue of current management integrity. Dieckamp had found the argument in the NOV response “kind of thin” but had chosen not to intervene. Id. at 334. The Commission characterized his behavior as “unwise,” but did not believe it raised a significant safety concern. Id. We cannot say that the Commission acted arbitrarily or capriciously in reaching these determinations.
Aamodts’ Radiation Release Study
Petitioners Norman and Marjorie Aamodt contend that the restart order was arbitrary and capricious because the Commission failed to reopen the record to consider further evidence concerning the health effects of the TMI-2 accident. The Aamodts first petitioned the Commission on June 21, 1984 for an investigation into the possibili*737ty that 1979 radiation releases were greater than reported by the licensee. Their petition included an empirical study, conducted by the Aamodts, of the health effects of the accident in three areas they believed fell in the pathway of radiation plumes released by TMI-2. This study purportedly documented flora abnormalities, an excessive number of cancer cases, and immediate post-accident physical responses symptomatic of radiation sickness.
The Commission requested and received a study of the Aamodts’ methodology by the Center for Disease Control (“CDC”), upon which it then relied, in a December 13, 1984 decision, to reject the request for further investigation. CLI-84-22, 20 NRC 1573 (1984). The Commission noted several problems with the Aamodts’ evidence. First, their conclusion flew in the face of the conclusions of numerous past radiation analyses. In addition, CDC’s analysis of the empirical evidence found that the Aamodts’ method of interview was unsystematic and subject to interviewer bias, that the Aamodts had failed to compare their findings to a control group, that the number of cancer deaths cited by the Aamodts was unverified, that in any event those cancer deaths had not been causally linked to the TMI-2 accident, and that the medically-accepted latency period from radiation exposure to cancer diagnosis was too long for cancer cases attributable to the TMI-2 accident to have surfaced. The Commission therefore held that the Aamodts had failed to present sufficient reliable information to discredit earlier, comprehensive, and scientific surveys of radiation releases from TMI-2. Finally, the Commission noted the existence of ongoing research into the health effects of the accident being conducted by the Pennsylvania Department of Health and promised to respond appropriately if this research produced any indication of high cancer risks to the surrounding population.
The Aamodts next moved for reconsideration and for reopening of the formal hearing record to receive the health information that they had compiled. They claimed to have verified the number, geographical placement, and medical cause of the cancer-related deaths described in their original motion. The Aamodts argued that the studies relied upon by the Commission in its December 13 decision fail to account for the fact that much of the data compiled in the first fifteen hours of the accident at TMI-2 are missing or for the possibility that radiation escaped, unmeasured by TMI-2 instruments, through “unmonitored release pathways.” The Aamodts also submitted their response to the CDC critique.
The Commission denied both motions on May 16, 1985. CLI-85-08, 21 NRC-, (1985). Applying the three factor test for reopening the record, the Commission found the motion untimely, not significant to safety, and unlikely to change previous results. In holding that the new evidence was not safety significant, the Commission continued to credit earlier studies over the Aamodts’, primarily because of problems in the time of diagnosis, assumed latency period, and population size in the Aamodts’ study. The Commission described the Aamodts’ evidence as “fragmentary and anecdotal” and insufficient “to support a reasonable doubt as to the adequacy and correctness of the several detailed scientifically conducted studies on which the Commission relied.” 21 NRC at-.
We do not find that the Commission’s refusal to reopen the record was arbitrary or capricious. The Commission found serious methodological flaws in the Aamodts’ study and it was not required to adopt their views regarding the statistical significance of the cancer cases they discovered or of the latency period between exposure to radiation and the occurrence of cancer. As for the propriety of the Commission relying on earlier studies of the health effects of the TMI-2 accident, we believe this is the kind of scientific determination over which “a reviewing Court must generally be at its most deferential.” Baltimore Gas & Electric Co. v. Natural Resources Defense Council, Inc., 462 US 87, 97, 103 S.Ct. 2246, 2252, 76 L.Ed.2d 437 (1983).
*738 Other Refusals to Reopen
Finally, the Commission refused to reopen the record on various other issues including changes in the so-called Lucien Report, a change in operator testimony at the Babcock and Wilcox trial, the allegedly untimely reporting of two consultant reports, and the alleged harassment of cleanup workers at TMI-2. In its fifty-seven page ppinion of February 25, 1985, the Commission addresses these issues in turn, setting forth the information available with respect to each, explaining the positions of the intervenors, and analyzing each in light of the three-prong test for determining when a record will be reopened. 21 NRC 282, 325-9, 335-42. No purpose would be served by our repeating that which is there published. Suffice it to say that the Commission’s decisions to pursue these matters only through staff investigation and not in a reopened adjudicatory hearing were permissible products of reasoned analysis.
VI. EMERGENCY PLANS
The Aamodts further contend that the NRC has failed adequately to respond to two other issues raised by them, to wit, that volunteer workers may not be relied upon to fulfill their evacuation-related responsibilities in the event of another accident at TMI, and that the Commonwealth’s emergency plan for farmers and their livestock is inadequate.
The Licensing Board held extensive hearings on emergency planning and, in particular, heard evidence concerning the reliability of volunteers. It directly addressed the issue in its second partial initial decision. LBP-81-59, 14 NRC 1211, 1486-89 (1981). The Board found “no evidence which contravenes the finding that there is reasonable assurance that in the event of a nuclear emergency at TMI there will be an adequate number of emergency workers who will stay and perform their jobs.” 14 NRC at 1489. We do not understand the Aamodts to contend that the Licensing Board’s decision finds no support in the record; rather their contention is that the Board was mistaken. As we have stressed earlier, however, our function is not to “second guess” the Commission’s resolution of the relevant issues.
With regard to the Commonwealth’s emergency plan for farmers and their livestock, the parties also extensively litigated this issue before the Licensing Board. 14 NRC 1211, 1671-80. The Board explicitly approved of the plan in its second partial initial decision. 14 NRC at 1677. The Appeal Board reached the same conclusion. ALAB-697, 16 NRC 1265, 1280 (October 22, 1982). Contrary to the Aamodts’ assertions, the Appeal Board did not order revisions of the plan that the Commonwealth has failed to effect. Rather, the Board found that the Commonwealth’s emergency plan goes “beyond the regulatory requirements and devotes considerable attention to the special needs of farmers,” id at 1275. The Appeal Board did, in fact, “strongly recommend” that “protective information specific to farmers be developed and distributed,” id at 1279, but it also clearly specified that its suggestions were not “a condition for restart.” Id. at 1280. We find no abuse of discretion in either of these decisions.
VII. DISQUALIFICATION OF THE LICENSING BOARD CHAIRMAN
Finally, TMIA contends that the Licensing Board Chairman, Judge Ivan Smith, prejudged the issues before him and demonstrated pervasive bias in favor of licensee, thereby depriving TMIA of its right to a fair hearing. The Commission refused to disqualify Judge Smith upon the motions of TMIA, the Commonwealth, and UCS. CLI-85-05, 21 NRC 566, April 5, 1985). Only TMIA presses this argument before us.
The standards governing recusal of NRC Licensing Board members are the same as those for the federal judiciary. Houston Lighting and Power Co. (South Texas Project, Units 1 and 2), CLI-82-9, 15 NRC 1363, 1365 (1982). A judge must disqualify himself if he “... has a personal bias or prejudice either against [a party] or in favor of any adverse party,” 28 U.S.C. *739§ 144. “The alleged bias and prejudice to be disqualifying must stem from an extrajudicial source and result in an opinion on the merits on some basis other than what the judge learned from his participation in the case.” United States v. Grinnell Corp., 384 U.S. 563, 583, 86 S.Ct. 1698, 1710, 16 L.Ed.2d 778 (1966). A rarely invoked exception to the extrajudicial source rule requires recusal when a judge displays “pervasive bias,” regardless of the source of the bias. Houston Lighting and Power Co. at 1366. Finally, a judge must also disqualify himself if “... his impartiality might reasonably be questioned,” 28 U.S.C. § 455(a). This test calls for recusal “where a reasonable man knowing all the circumstances would harbor doubts concerning the judge’s impartiality.” United States v. Dalfonso, 707 F.2d 757, 760 (3d Cir.1983).
TMIA points to several examples of allegedly biased courtroom behavior on the part of Judge Smith. We have reviewed each of the designated episodes. In context, they simply do not support the petitioners’ claim of personal bias in favor of GPU.
Of more substance is TMIA’s contention that a letter, written by Judge Smith to District Court Judge Sylvia Rambo, both represented prejudgment of the issues before the Licensing Board and stemmed from extrajudicial sources. This letter urged Judge Rambo to be lenient in her sentencing of a TMI-2 operator convicted of cheating on licensing tests. Although we believe Judge Smith exercised poor judgment in writing this letter, we do not find that, taken alone or in conjunction with the other TMIA examples of bias, the letter requires disqualification of Judge Smith. Judge Smith’s letter provided three reasons for leniency, only one of which arguably raises a question of prejudgment of the merits in the Restart Proceedings. The letter stated:
[A] severe criminal penalty ... is in my personal view, not needed to insure the integrity of the NRC operators’ licensing process at Three Mile Island____ Many weeks of public NRC hearings have been devoted to the issues of TMI management integrity and operator competence and, in fact, hearings on that very issue are still in progress. I have confidence that the NRC administrative regulatory process, with extensive public participation, will provide an orderly and reliable mechanism for assuring that any problems caused by deception respecting Three Mile Island will have been identified and resolved. Deception in the future is very unlikely.
Although this argument may be said to express confidence in the NRC rather than bias in favor of GPU, it appears to express confidence that the NRC will develop procedures that the Licensing Board will find “necessary and sufficient” to protect the public health and safety. This is, of course, closely related to what the Commission, in its August 9, 1979 Order, directed the Board to determine, with one important difference. Judge Smith did not indicate whether specific training-related actions would be sufficient to support restart. Rather, he stated that the NRC, with public participation, would eventually develop a “mechanism” for preventing future deception. Thus, Judge Smith did not prejudge the necessity or sufficiency of any of the short- or long-term actions that the Commission charged the Licensing Board to evaluate. He simply indicated his belief that any future deception would not go undetected.
We find no fault with the Commission’s refusal to disqualify Judge Smith.
CONCLUSION
The 1979 accident at Three Mile Island presented the NRC with a substantial and difficult problem concerning the continued operation of the undamaged reactor, TMI-1. The Commission ordered that the reactor remain in a shutdown condition pending investigation and an adjudicatory hearing in which interested representatives of the public could participate. As the investigation and the hearing progressed, it became increasingly apparent that there had been serious deficiencies in the licensee’s performance at TMI-2. This *740fact made the Commission’s task of resolving the TMI-1 restart issues considerably more involved and resulted in the Restart Proceedings being significantly more comprehensive than originally anticipated. The procedures chosen by the Commission well served their intended purpose, however, and a wide variety of issues, including the competence of the licensee’s management, were aired and resolved. The intervenors played an active and constructive role in those proceedings. With their help, 155 conditions were fashioned that the Commission has determined provide adequate assurance that TMI-1 can be operated by the licensee with safety.
Petitioners are not satisfied with the Commission’s process or with the results of its deliberations primarily because, after the hearing, they learned, or first focused upon, certain information that they deem relevant to the management competence issues. While they were permitted to bring this information before the Commission and secured a staff investigation concerning it, they were unsuccessful in their efforts to reopen the adjudicatory hearing. They feel strongly about the issues raised in their motions to reopen and their desire for further hearings is understandable. At some point, however, proceedings must terminate in outcomes. The Commission determined that this point has been reached with respect to the issue of lifting the 1979 shutdown orders, and we conclude that this determination is neither arbitrary nor capricious.
When Congress enacted the Atomic Energy Act, it was aware that risk assessment and policy choices would be an inherent part of the development of atomic energy. It assigned the responsibility for making such assessments and choices to the NRC. Insofar as the issues presented by these petitions for review are concerned, the NRC has exercised its authority reasonably and in accordance with law. Accordingly, we will deny the petitions for review.
. TMI-1 and TMI-2 are jointly owned by Metropolitan Edison Company, Jersey Central Power & Light Company, and Pennsylvania Electric Company. Metropolitan Edison Company was the NRC-licensed operator of TMI-1 and TMI-2 until January 1982. Since then, GPU Nuclear Corporation has been the licensed operator of both units. All of these companies are wholly-owned subsidiaries of General Public Utilities Corporation.
. The original Licensing Board members were Ivan W. Smith, Esquire, Chairman, Dr. Walter H. Jordan, and Dr. Linda W. Little. 10 NRC at 146-7.
. The Commission originally planned to conduct the appellate review of the Licensing Board's decision itself. However, the complexity of the proceedings led the Commission to establish an Atomic Safety and Licensing Appeal Board (“Appeal Board”) to hear initial appeals. CLI-81-19, 14 NRC 304 (1981).
. Ordinarily, a licensee is given the opportunity to challenge an NRC enforcement action before the action takes effect. Because the Commission determined that protection of public health and safety required issuance of immediately effective shutdown orders for TMI-1, it did not give Metropolitan Edison this opportunity. Therefore, the Commission provided for restart, prior to final appellate review of Licensing Board decisions, once the reasons for making the order immediately effective were adequately resolved. CLI-79-8, 10 NRC 141, 149 (1979).
. Shortly before the Licensing Board issued its first partial decision, the NRC staff notified the Board that it was investigating allegations of cheating by licensee’s senior reactor operators on NRC operator license exams.
. The Commission’s Order of August 9 directed the Licensing Board to determine whether TMI-1 may be safely operated while TMI-2 was being decontaminated. 10 NRC at 145.
. See supra footnote 1. Besides managing TMI-1 and TMI-2, GPUN operates the Oyster Creek nuclear facility.
. After conducting an investigation, the staff once again endorsed management’s competence in July, 1984. NUREG-0680, Supp. No. 5 (July, 1984).
. Dieckamp remained on the GPUN Board of Directors.
. The Commission also exempted from the proceeding employees already exonerated by the NRC’s Office of Investigations, which earlier had investigated allegations of leak rate falsification at TMI-1. 21 NRC at 287.
. Unless otherwise noted, the cited regulations remain unchanged since 1979.
. We are not called upon to determine, and we do not determine, whether, in fact, the immediate restart of TMI-1 with the conditions imposed by the Commission would be incompatible with the requirement that the facility operate "in accordance with” its existing operating license. 42 U.S.C. §■ 2131.
. We are mindful that the Commission claims the authority to effect license amendments during an enforcement proceeding upon a finding that the public health and safety require that such amendments become immediately effec*730tive. Nothing suggests to us that it has exercised that authority in this case. Under 42 USC § 2131, it is unlawful for anyone to use a "production facility except under and in accordance with a license issued by the Commission.” For obvious reasons, it is important that the parameters of a licensee’s operating authority be clear and unambiguous. We think it highly unlikely that the Commission would attempt to amend a license without expressly stating its intention to do so and without setting forth the amendment in haec verba. It seemed apparent at oral argument that the issue of which “conditions" of the lifting of the shutdown order would become license amendments had not as yet been addressed.
. Petitioners also argue that the Administrative Procedure Act (APA) requires that they be afforded full hearing rights on all material issues in the Restart Proceedings, and that the Commission’s decision be based on the adjudicatory record. The APA specifies the procedure to be followed “in every case of adjudication required by statute to be determined on the record after opportunity for an agency hearing,” 5 U.S.C. § 554 (emphasis added). Petitioners’ claim to APA rights consequently depends on the applicability of § 189(a) to this case. Since we do not reach the question of § 189(a)’s applicability, we also have no comment on the merits of petitioners’ APA arguments.
. The August 9, 1979 Order expressly contemplated a Commission decision on whether to lift the shutdown orders prior to completion of the appellate review process. CLI-79-8, 10 NRC 141, 149 (1979). The Appeal Board, after its creation, was powerless to lift the stay. Rather, the Commission was "the exclusive administrative body with the power to determine whether *732Unit One may restart during the pendency of any possible appeals of a Board decision before the” Appeal Board. CLI-81-34, 14 NRC 1097, 1098 (Dec. 23, 1981).
. Strictly speaking this issue is not "new.” Allegations of leak rate-data manipulation were made by a TMI-2 operator, Harold Hartman, as early as May 22, 1979. In fact the “Hartman allegations” received New York television coverage in March, 1980. Nevertheless, because the NRC did not actively pursue the matter and may have at least implicitly discouraged petitioners from doing so, in light of the ongoing criminal investigation, see ALAB-738, 18 N.R.C. 177, 188 (1983), the Commission did not rest its decision on the timeliness of the motions to reopen. The fact that petitioners may not have been guilty of laches in failing to earlier press the Hartman allegations, however, does not mean that their motions to reopen the record should be tested by a lower standard than if they were asking for consideration of newly discovered evidence.
. A fourth employee associated with TMI-2 is Brian Mehler, now manager of the TMI-1 Rad-waste Department. The Commission noted that Mehler no longer has direct involvement with the operation of the TMI-1 reactor, 21 NRC at 298-99, and no petitioner seriously challenges the safety significance of his continued presence.