Opinion for the Court filed by Senior Circuit Judge FLOYD R. GIBSON.
Concurring opinion filed by Circuit Judge SILBERMAN.
FLOYD R. GIBSON, Senior Circuit Judge:Petitioners Consolidated Edison Company of New York, Inc., Duquesne Light Company, and Ohio Edison Company (“the utilities”) challenge a portion of a final rule promulgated by the Department of Energy (“DOE”) pursuant to the Nuclear Waste Policy Act, 42 U.S.C. §§ 10101 et seq. (1982) (“NWPA” or “the Act”).
The rule provides that utilities that generate nuclear power must pay a statutorily prescribed fee based on “net kilowatt hours generated.” 52 Fed.Reg. 35,356, 35,359 (1987). The utilities claim that the rule violates the plain language of the NWPA by assessing the fee on electricity that is not actually sold. We agree and thus grant the utilities’ petition for review.
I.
In the NWPA, Congress charged DOE with the responsibility of administering a program for the permanent disposal of spent nuclear fuel and high-level radioactive waste produced by civilian nuclear power plants. The costs of the disposal program, according to the Act, are to be borne by electric utilities that generate nuclear power and consequently produce nuclear waste. Subsection 10222(a)(2) is part of the NWPA’s system of levying user fees on the utilities that generate nuclear power and provides that:
For electricity generated by a civilian nuclear power reactor and sold on or after the date 90 days after January 7, 1983, the fee ... shall be equal to 1.0 mil [sic] per kilowatt-hour.
42 U.S.C. § 10222(a)(2) (1982).
In April 1983, DOE adopted a final rule providing that the ongoing one mill fee under subsection 10222(a)(2) would be charged on all electricity generated by each nuclear power plant. 48 Fed.Reg. 16,590, 16,602 (1983). A group of electric utilities, including two of the three utilities involved in this case, challenged that rule in Wisconsin Electric Power Co. v. Department of Energy, 778 F.2d 1 (D.C.Cir.1985) (“WEPCO”). The utilities in WEPCO argued that DOE may not charge the one mill fee for electricity used by a utility on-site to operate its plant because that electricity is not “generated ... and sold” as required by subsection 10222(a)(2) of the Act. Id. at 3. We agreed with the utilities in WEPCO and thus granted their petition for review. We found that the “generated ... and sold” language plainly and unambiguously requires that the fee under subsection 10222(a)(2) be assessed only on electricity that is both generated and actually sold. Id. at 8.
In September 1987, after having published a Notice of Proposed Rulemaking, 51 Fed.Reg. 40,684 (proposed Nov. 7, 1986), and receiving comments thereon, DOE promulgated the final rule that is at issue in this case which reflects DOE’s attempt to bring its fee into compliance with the Act in light of our holding in WEPCO. 52 Fed.Reg. 35,356 (1987). The rule provides for a one mill assessment on “net kilowatt hours generated,” which is defined as “gross electrical output produced by a civilian nuclear power reactor measured at the output terminals of the turbine generator minus the normal onsite nuclear station service loads during the time electricity is being generated.” Id. at 35,359-60.
A number of interested parties submitted comments on the rule as proposed and contended that DOE’s definition of “net kilowatt hours generated” failed, like the final rule rejected in WEPCO, to adequately reflect the “generated ... and sold” language of the Act. Commenters argued *696that although DOE’s definition of “net kilowatt hours generated” excluded electricity used on-site, it failed to exclude electricity lost in the transmission and distribution (“T & D”) process and electricity used by the utilities to operate their off-site facilities. Commenters suggested two methods of deducting T & D losses in calculating electricity generated and sold. It was suggested that DOE adopt a uniform deduction of 8.65 percent for all utilities to account for T & D losses. This figure is based on an industry-wide average of kilowatt hours not sold. Alternatively, commenters suggested that DOE assess the one mill fee on the basis of a utility-specific calculation of the percentage of electricity generated that is nuclear multiplied by the total number of kilowatt hours sold to paying customers as measured at the customers’ meters. Final Brief for Petitioners at 8-9.
DOE rejected the argument that T & D losses and electricity used to operate off-site facilities should be excluded from the calculation of electricity generated and sold. It stated in its final rule that its new approach “follows the guidance of [WEP-CO ] and ensures that, as envisioned by Congress, current ratepayers carry their fair share of the burden of paying for the disposal of nuclear waste.” 52 Fed.Reg. at 35,357. DOE specifically rejected the com-menters’ proposal of a 8.65 percent uniform reduction to account for T & D losses, reasoning that that method “would result in substantial inequities among utilities.” Id.1 DOE also stated in its final rule that it examined several utility-specific approaches to calculating T & D losses but rejected them because the additional data they would require from utilities “would substantially increase the administrative burden of the utilities as well as that of DOE.” Id.
This petition for review followed.
II.
In WEPCO, this circuit decided that because subsection 10222(a)(2)’s “generated ... and sold” language plainly and unambiguously provides that the fee applies only to electricity that is both generated and actually sold, DOE may not charge the fee for electricity that a generating plant itself consumes. We rejected DOE’s argument that strict adherence to the express terms of the statute would produce the inequitable result of future ratepayers paying for the disposal of earlier users’ nuclear waste and that this result contravenes Congress’s stated, overriding intent. WEPCO, 778 F.2d at 6. We were unconvinced that Congress had manifested an intention, either in the Act itself or in its legislative history, that the “and sold” provision should be read in any fashion other than in its plain meaning. Id. at n. 11. DOE invoked, and we fully accepted, the general proposition that “an agency’s interpretation of the statute it has been directed to administer is entitled to ‘great deference.’ ” Id. at 8 (quoting Appellees’ Brief at 23 and Udall v. Tallman, 380 U.S. 1, 16, 85 S.Ct. 792, 801, 13 L.Ed.2d 616 (1965)) (footnote omitted). However, we answered DOE’s call for deference with the irrefutable principle that deference is not accorded to any agency interpretation that overrides the clearly expressed intent of Congress. Id. The crux of our decision in WEPCO was our determination that the statutory language at issue was clear and unambiguous. Having found that congressional intent with respect to the disputed language was clear, we necessarily concluded that the statute’s plain meaning must be given effect because neither an agency administering a statute nor a court interpreting it is at liberty to override Congress’s clearly expressed intent. Id. See also Chevron U.S.A. Inc. v. Natural Resources Defense Council, Inc., 467 U.S. 837, 842-43, 104 S.Ct. 2778, 2781-82, 81 L.Ed.2d 694 (1984) (stating that “[i]f the intent of Congress is clear, that is the end of the matter; for the court, as well as the agency, must give *697effect to the unambiguously expressed intent of Congress”) (footnote omitted).
The principles that guided us in WEPCO also control in this case. Electricity that is lost in the T & D process and electricity used to run off-site facilities, like electricity that is consumed on-site, is not electricity that is “generated ... and sold” within the plain meaning of subsection 10222(a)(2). Thus, we agree with the utilities that in order to bring its fee into compliance with the Act DOE must adopt a reasonable method of deducting from the amount of electricity upon which it assesses the one mill fee the amount of electricity lost in the T & D process and the amount of electricity that is used to operate off-site facilities.
III.
DOE asserts in its final rule that it has not disregarded this court’s decision in WEPCO. DOE reasons that since the WEPCO petitioners chose not to assert the point that “and sold” should exclude T & D losses and electricity used off-site, see WEPCO, 778 F.2d at 5 n. 10, “these losses were not addressed in the [WEPCO court’s] interpretation of the term ‘generated * * * and sold’ as used in the Act.” 52 Fed.Reg. at 35,357.
We recognize that the issue of T & D losses and electricity used off-site was not raised in WEPCO. But nothing in our WEPCO decision, explicitly or otherwise, suggests that our conclusion about the plain meaning of “and sold” applied exclusively to electricity consumed on-site. We held in WEPCO that effect must be given to Congress’s clear intention that the fee be levied only on electricity that is both generated and actually sold. Today, we are bound by that holding.2
As further support for its rule, DOE suggests that we should not cause further reductions in the revenues it collects from the ongoing fee because in WEPCO this court “was concerned that DOE adopt a ‘modest’ approach in defining net generation of electricity so as not to defeat Congress’ intent that an unfair burden not be placed on future ratepayers if DOE initially collects insufficient revenues.” 52 Fed. Reg. at 35,357. This “inequity to future ratepayers” theory was DOE’s primary argument in WEPCO and we stated in response that we were “unable to find any inconsistency with Congress’ intent that would be created by strict fidelity to the express terms of the statute....” WEPCO, 778 F.2d at 6 (footnote omitted). We are compelled to reach the same conclusion here.
We acknowledge in this regard that our decision today widens the “gap” between the total amount of electricity generated by. the nuclear power plants and the amount of that electricity upon which the ongoing fee. may be assessed. In WEPCO we found it significant that exclusion from the fee of electricity used on-site resulted in a “modest” five percent reduction in the amount of electricity to which DOE may apply the fee. Id. at 8. Deducting electricity lost in the T & D process and electricity used off-site will cause a further reduction of approximately 8.65 percent, or perhaps, somewhat less. Nevertheless, we remain convinced that Congress plainly intended that the ongoing fee applies only to generated electricity that is actually sold, and neither this court nor DOE is at liberty to depart from Congress’s dictate. Further, we note as we did in WEPCO that DOE is not without a remedy in the event that it finds that insufficient revenues are being collected to pay for the disposal program. Under subsection 10222(a)(4) of the Act, the Secretary is directed to review annually the level of the ongoing fee in order to determine whether sufficient revenues are *698being collected to offset the costs of administering the program. If the Secretary determines that insufficient revenues are being collected, according to subsection 10222(a)(4), he is directed to “propose an adjustment to the fee to insure [sic] full cost recovery.” 42 U.S.C. § 10222(a)(4) (1982) (footnote omitted). This annual review-and-adjustment mechanism buttresses our conclusion that the result we reach today does not thwart Congress’s intent by unfairly burdening future ratepayers. Congress provided in subsection 10222(a)(4) a means to correct the problem which DOE anticipates. In any event, we are not at liberty to depart from Congress’s clearly expressed intention.
IV.
DOE has argued, and we accept, that it is difficult, if not impossible, to measure with precision the amount of electricity lost to the T & D process and off-site use. This difficulty, however, does not relieve DOE of the obligation to implement some reasonable and fair method of accounting for these losses in charging its fee. We fully recognize that to some extent DOE must rely on approximations or industry-based averages in deducting the amount of electricity used off-site and lost in the T & D process, and we leave it to the agency to decide upon a reasonable method. But, despite the obstacles to precise measurement, we require that DOE adopt a method that implements Congress’s intent that the ongoing fee under subsection 10222(a)(2) applies only to electricity that is both generated and actually sold.
For the reasons stated, the petition for review is granted.
SO ORDERED.
. DOE also disagreed with the commenters’ conclusion that 8.65 percent represents the industry-wide average of kilowatt hours not sold due to T & D losses. DOE stated in its final rule that reports to the Federal Energy Regulatory Commission show an average 6.9 percent value of electricity not sold due to T & D losses. 52 Fed.Reg. at 35,357.
. DOE does not identify, and we are not aware of, any doctrine of waiver or estoppel that applies in these circumstances to preclude the utilities, two of which were petitioners in WEP-CO, from asserting the points not raised in WEPCO that T & D losses and electricity used off-site must be excluded from the ongoing fee. See City of Cleveland v. Federal Power Comm’n, 561 F.2d 344, 347-48 (D.C.Cir.1977) (vacatur of order on basis of one cited inconsistency between order and controlling legal principle does not preclude parties from citing further exemplary inconsistencies during subsequent proceedings on remand).