OPINION OF THE COURT
Jones, J.We conclude that there was a rational basis for and substantial evidence in the record to support the determination of the Public Service Commission authorizing the introduction of time-of-day pricing as the basis for a new rate structure for the furnishing of electricity, and, as a first step to that end, approving the proposal of Long Island Lighting *665Company to initiate the program by charging particularly designed time-of-day rates to a small group of its largest commercial and industrial customers.
On August 8, 1975 Long Island Lighting Company (LILCO) filed an application with the Public Service Commission for a general rate increase. In response the commission issued an order initiating case No. 26887, which order also directed the company to file "rate proposals to price electricity by time of day”. Pursuant to that directive LILCO filed proposed rate "Service Classification 2 — Multiple Rating Period” (hereafter "SC2-MRP”). That filing was served on all parties to case No. 26887 and as well on all parties to case No. 26806 (proceeding on motion of the commission as to rate design for electrical corporations).
On January 29, 1975 the commission had issued an order stating that the "rapidly increasing costs of new generating facilities and the rising cost of fuel both make it urgent, in the interest of energy conservation and the efficient use of resources, that the structure of energy prices reflect, to the greatest extent feasible, the variations in the incremental costs of service because of differences in the time of consumption, as well as in all other cost-influencing factors”. By the same order the commission instituted case No. 26806, the so-called "generic proceeding”, to consider, among other things, "whether marginal or incremental costs provide a reasonable basis for determining the rate structure of electric utilities”, and to resolve certain issues concerning the legality, theory and practicality of marginal cost pricing before approval of any specific rate proposals. On the basis of an extended record (which included the testimony of a host of economists, engineers and experts in rate design who had been subjected to cross-examination by the Council of Retail Merchants and also included the testimony of an economist who appeared on behalf of the council) the commission concluded in that proceeding on August 10, 1976 "that marginal costs do provide a reasonable basis for electric rate structures”. The commission added, however: "This finding does not mean that rate structures must in all cases embody marginal cost pricing, or that rate structures in any case should be based exclusively on such principles. But it does mean that marginal costs are an important tool for consideration in all rate cases, and that failures to take these principles into account should be justified”. After noting its cognizance of the problems of both the *666utility companies and their customers and denying any intention to inflict unnecessary or unreasonable hardship, the commission stated: "To the extent that implementation of marginal principles might involve abrupt changes, the Commission’s guiding principle will be gradualism consistent with appreciable improvement.” Noting that LILCO and Consolidated Edison Company of New York had already presented marginal cost analyses, other electric utilities were directed to transmit their studies, and the commission advised that it would proceed to evaluate LILCO’s proposed marginal cost-based rates in case No. 26887. No aspect of case No. 26806, which has not been concluded, is now being appealed.
Accompanying LILCO’s filed proposed rate SC2-MRP was the marginal cost study which formed the basis of the proposed rate. LILCO submitted expert testimony with respect both to the manner in which marginal costs were translated into rates and to the economic basis for the proposed SC2MRP rate. This testimony was subject to cross-examination by the Council of Retail Merchants which also introduced expert testimony of its own.
On December 16, 1976 the commission approved SC2-MRP. The council’s request for a rehearing before the commission was denied. The council then instituted the present proceeding under CPLR article 78 which was transferred to the Appellate Division. On May 31, 1978 that court annulled the commission’s determination. The commission and LILCO have appealed to our court. We now reverse the judgment of the Appellate Division and confirm the order of the commission.
As a matter of analysis it appears that our consideration should proceed on three levels. The first involves the validity of time-of-day rate structures in principle; the second, the validity of the particular time-of-day rate structure proposed by LILCO; and the third, the permissibility of the classification of consumers to which LILCO proposes initially to apply its proposed time-of-day rate structure.
As to the first level there is no dispute. Respondents do not challenge the validity in principle of a rate structure based on quantity and time of consumption. Indeed classification of service "based upon the quantity used [and], the time when used” is expressly authorized by statute (Public Service Law, § 66, subd 14; cf. id., § 65, subd 5).
Because the Appellate Division annulled the commission’s determination at the third level, concluding that the proposed *667SC2-MRP rate "constitutes an unlawful inter-class price discrimination in violation of subdivisions 2 and 3 of section 65 of the Public Service Law” (62 AD2d, p 316), without passing on the validity of the time-of-day rate structure itself, we first address this aspect of the case on which the disposition below was based. The promulgation of any classification of consumers, any subdivision of the whole, necessarily, of course, involves differentiation; the critical question is whether the particular classification works an undue or impermissible discrimination. It appears that the Appellate Division took the view that to withstand assault the particular classification must be based on some identifiable cost-justification, that rate fixing that departs from cost-justification would produce an undue preference or advantage favoring those who are not within the class, in violation of the provisions of subdivision 3 of section 65 of the Public Service Law. That court then rejected the commission’s and LILCO’s attempt to demonstrate that there was a direct cost allocation basis to support the proposed consumer classification. While not abandoning their contentions that cost-justification can be shown in support of the proposed classification the commission and LILCO contend, and we agree with them, that in this instance at least the choice of consumers to be included within SC2-MRP was made on a rational basis incident to the phasing in of a new rate structure, and that selective, step-by-step implementation of a new theory of rate fixing is permissible.
LILCO, having been directed by the commission to file a rate based on marginal costs, with the benefit of suggestions made by commission staff, calculated the company’s marginal costs and then placed those costs into three rating periods. The choice of the particular periods was a function of metering capability, homogeneity of costs within different hours, the probability of loss of load for each hour and the likelihood of consumer reaction. As a result of this weighing process the following periods were identified:
Period 1 (off-peak period, lowest demand): Midnight to 7 a.m.; all days, all year;
Period 2 (peak period, greatest demand): 10 a.m. to 10 p.m. weekdays and Saturdays, June 1 to September 30; and
Period 3 (intermediate or "shoulder” period): all remaining times.
The company then selected the group of its consumers to which new rates based on quantity and time-of-day consump*668tion would be applied in the first phase of eventual across-the-board application. It chose its largest commercial and industrial consumers. The line of demarcation was fixed at those consumers whose peak demand exceeded 750 kilowatts in any two of the preceding 12 months — as of September, 1976 there were about 175 customers in this category.
The commission’s interest in the development of a more sophisticated rate structure was predicated on its view that the present condition of the State’s economy made such development imperative. "Repeatedly in recent rate decisions we have stated our intention to hold rates down to the lowest reasonable level, precisely because we realize that energy costs have become oppressive to large numbers of consumers and have a magnification effect throughout the economy. Over two years ago we embarked on a program of utility management and efficiency audits and, additionally, we are demanding commitments to increased productivity. We are confident that these efforts will help to hold rates down; but they alone cannot reach the two principal cost factors that have pushed utility rates to unprecedented levels: the escalating costs of new construction and of fuel, particularly oil.” Prior to the present proceedings, time-of-day and marginal cost pricing principles had never been used in New York State as a basis for pricing electricity. In its consideration of this new approach to rate fixing the commission recognized that peak usage by every customer imposes the same costs on the system and did not seriously contend that a time-of-day rate structure would not affect all of LILCO’s customers approximately evenly, conferring similar benefits and imposing comparable burdens.
In the implementation of the new rate structure by LILCO, however, the commission made two critical determinations. First, it determined that there should be "selective implementation as a part of a programmed plan”, that the "interest of all ratepayers is best served by a step-by-step application of time-of-day metering because of the enormous cost that a widescale mandatory metering program would entail”. Thus, it rejected contentions that time-of-day metering should initially be applied across the board. Second, acknowledging that "[a]ny cut off point for the first step inevitably involves some element of arbitrariness, as does any reasonable classification”, it approved the particular classification of consumers proposed by LILCO for its first phase implementation.
*669 We have held that rate discrimination can be countenanced only if it is either cost-justified or if some other rational basis is to be found in the record. (Matter of Lefkowitz v Public Serv. Comm., 40 NY2d 1047, 1048.) Thus, it is not conclusive that there may not have been shown a sufficient cost-justification for the proposed classification or even that it departed from cost allocation, if there exists another rational basis for that classification.1 In our view the commission’s determination in approving LILCO’s proposal must be upheld if there exists a rational, though not necessarily cost-related, predicate for the commission’s action. We find that there is a rational basis both for the determination that introduction of the newly designed rate structure should proceed on a step-by-step basis rather than across the board at the outset, and for the conclusion that the first step classification proposed by LILCO was reasonable in the circumstances.
It appears that there had been no consumer or utility experience with the new rate structure in theory or in practice, that its implementation will be possible only with the installation of relatively expensive metering devices and that the regulatory body has found that effective utilization will "require detailed and individual consumer education on an individual basis” which "can be done most efficiently with a relatively small group of consumers at the outset”. On such a record we cannot say that the courts are entitled to substitute their judgment for the evaluation of the commission, giving fair consideration to the expertise possessed by the commis*670sion in weighing the impact of rate-fixing factors on both the utilities and the consuming public and reaching the ultimate determination as to how appropriately to proceed in this given instance. To the extent that it is applicable, equal protection analysis recognizes the right of government to move forward by gradual progression rather than by universal application or not at all. (Erznoznik v City of Jacksonville, 422 US 205, 215; Dandridge v Williams, 397 US 471, 486-487; Williamson v Lee Opt. Co., 348 US 483, 489.) Thus, we conclude that there was warrant for the commission’s determination to implement time-of-day rates in progressive steps rather than to apply the new rates to all customers at one time.
By similar analysis we find that the record compels the conclusion that there was a rational basis for approval of the group of consumers selected by LILCO for first stage application of the new rate structure. Several considerations entered into the choice of the class of consumers to which the new rates would be applied in the first instance. Prior to having been required to propose a marginal cost-based rate, LILCO had at its own expense placed the type of sophisticated meters required by such a rate with its largest use commercial and industrial class of customers for the purpose of collecting load data. Not only could it be found that it was appropriate to use the large use consumers because their high kilowatt hour usage would result in a much lower metering cost per kilowatt hour, thus producing a minimal customer burden relative to what it would be for other customers on LILCO’s system; additionally, meters were already in place for these consumers, thereby postponing "the enormous cost that a widescale mandatory metering program would entail”. Necessary consumer education can most efficiently be undertaken with a relatively small group of informed, sophisticated consumers. The group selected was making substantial payments to LILCO and thus had a real potential for usage responsiveness; its large consumption of energy offered both opportunity and inducement to take effective action, perhaps even at some initial cost to the consumers, to shift more load to off-peak periods. There was warrant for the hope that some significant decrease in the use of inefficient generating facilities might be realized, even without proof as to the precise degree of elasticity in energy consumption among the included consumers. Here again the identification of relevant factors, the weighing of the relative significance of each and of all taken in combi*671nation, the appropriateness of the balance struck in the ultimate selection of the group of customers for initial application of the new rate structure — all are matters lying within the areas of the commission’s expertise with respect to which the courts should be realistically reluctant to intervene.
In passing we note, that the commission considered and rejected the suggestion that the new rate structure should first be put into effect on a voluntary rather than a mandatory basis. The fact that such a plan of implementation has been adopted in other States and surely could be said to have a rational basis, does not establish that it is the only rational plan or that the mandatory plan approved by the commission in this instance did not have a rational basis. Indeed, other considerations of public acceptance and of the effect of consumer self-selection might raise questions as to the efficacy and desirability of introduction of time-of-day rate fixing in any voluntary manner. At the very least there were matters lying well within the scope of the commission’s authority, out of reach of judicial intrusion. Then, too, we observe that in the design of the particular first phase implementation, efforts were made to assure that the selected consumers as a class will not pay total revenues to LILCO greater than they would have had to pay had their rates not been redesigned.2 Also steps were taken, for instance by the application of the so-called "two-thirds rule”3 to moderate the effect of the new rates. These factors, too, bear on the rationality and reasonableness of the particular implementation plan which was approved by the commission.
In sum we conclude that the record compels the conclusion that there was a rational basis for the determination of the commission to approve both the introduction of time-of-day rate fixing on a gradual progression basis and the first stage application proposed in this instance by LILCO. Thus, we disagree with the conclusion reached at the Appellate Division.
It remains then to consider whether the commission’s adoption and calculation of the SC2-MRP rate is supported by the evidence in the record — the issue which was not reached *672at the Appellate Division. We reject respondents’ arguments that the commission’s approval of SC2-MRP was both unreasonable and unsupported by substantial record evidence. At the outset we observe again that we are in an area presenting problems of a highly technical nature, the solutions to which in general have been left by the Legislature to the expertise of the Public Service Commission. As the commission observed, "judgment has had to be exercised in assessing costs and translating them into rates”. It is only when it can be shown that the exercise of judgment was without any rational basis or without any reasonable support in the record that the determination of the commission may be set aside (Matter of Campo Corp. v Feinberg, 279 App Div 302, 307, affd 303 NY 995). "The courts will not interfere with the conclusion [of the rate-making agency] except to safeguard the consumer against arbitrary power” (City of Rochester v Rochester Gas & Elec. Corp., 233 NY 39, 49; Permian Basin Area Rate Cases, 390 US 747, 797-799).
LILCO’s SC2-MRP classification was derived from a marginal cost study which focused on three principal areas of cost: (a) marginal energy or running costs; Ob) marginal capacity costs; and (c) marginal customer costs.4 There was uncontroverted evidence that production costs increased in times of peak demand when it became necessary to put the less efficient, more costly generating facilities on the line. Thus, in this aspect the marginal running cost of producing electrical energy is related to the costs associated with the utilization of the least efficient generating plant.
Respondents specifically attack the component of marginal capacity cost — costs associated with the construction of new plant. LILCO used long-run marginal capacity costs in conjunction with the short-run marginal running costs. There is expert testimony in the record that this combination, as a practical matter, yielded the most economically efficient price at the time, and that in this instance inasmuch as curtailment costs (the value of the output which would not be produced if electricity demand exceeded supply, causing blackout or brownout) could not be computed; long-run incremental capacity costs were a reasonable surrogate for short-run curtailment costs. These were matters to be judged by the commission.
*673Similarly respondents challenge the use of projected costs of a prospective gas turbine generator which it was said would be purchased to meet peak demand if such demand approached LILCO’s system generating capacity, on the ground that LILCO had determined that it presently had more than ample capacity to meet any possible demand for many years hence, and thus that any such approach is totally hypothetical. How to balance such arguments against countervailing contentions is properly to be left to commission expertise and judgment. Thus, it was for the commission to weigh assertions that inasmuch as generating capacity is added in lumps it is necessary to consider fairly large increments in forecast demand, and that, even if there be present overcapacity which is likely to continue to exist for some time, recognition must nonetheless be given the long lead time required for the construction of new generating facilities. It was for the commission, in the light of such contentions, to determine whether and to what extent LILCO’s scheduled installation of five gas turbines in 1976 (although later canceled when the level of projected peak demand did not materialize) provided a relevant basis for computing incremental generating costs. Indeed on the record before the commission it appears that consideration of this factor was not opposed and that, as the commission recited, "there was little controversy in this case about the use of the combustion turbine as the measure of marginal capacity cost”. Likewise the estimation of LILCO’s "loss of load probability” (the risk of demand exceeding supply) in the face of LILCO’s recognized surplus capacity and the evaluation of its impact on marginal capacity cost involved the selection and weighing of operational and economic factors properly to be left to the commission.
Finally, say respondents, a primary goal of marginal cost-based pricing is to conserve resources and to improve utility efficiency (in theory by sending consumers price "signals” which are designed to influence their consumption patterns), but, respondents continue, the record does not support the conclusion that this will be the result in this instance inasmuch as the initial group of consumers was selected without regard to their ability to respond to price signals. The determinative fact, however, is that the record does not establish that conservation of resources and improved utility efficiency will not be the result of the new rate structure. It suffices if there is evidence that they may be. Even in the *674absence of proof that each of the affected consumers had an elasticity of demand in consequence of which it could respond to price signals, it cannot be concluded that the new rate structure will not produce significant changes in consumption patterns, whether by way of shifts from peak period consumption to consumption in off-peak or shoulder periods or by way of reduced consumption of energy in consequence of the installation of more efficient electricity using equipment or the redesign of present equipment to achieve the same end. It will only be after the new rates have been in operation that the practical effect of the carrot and the club can reliably be measured. The determination of the commission could be set aside on this ground only if it were demonstrated on the record that the new rate structure would not operate to conserve resources or to improve utility efficiency. The record in this case would support no such adverse finding. Additionally the real test on this approach will come when time-of-day pricing is applied across the board, not alone to the small segment of consumers to which the new rate structure is initially applied in first stage implementation, but to all of LILCO’s customers. We cannot say, with the evidence which is in the record that the pricing signals contained in the SC2MRP tariff might not be effective to achieve the desired goal or that there is insufficient warrant for the commission’s determination to take the first step into time-of-day pricing and to authorize LILCO’s selective implementation of the program starting with the largest commercial and industrial consumers.
Recognizing the novel character of time-of-day pricing for electricity we conclude that the determination of the commission in approving LILCO’s SC2-MRP rate proposal represents a rational and reasonable step in the direction pointed by the commission toward time-of-day pricing for electricity and that it finds substantial support in this record.
For the reasons stated the judgment of the Appellate Division should be reversed, with costs, and the determination of the commission confirmed.
Chief Judge Breitel and Judges Jasen, Gabrielli, Wachtler, Fuchsberg and Cooke concur.
Judgment reversed, etc.
. That classification approval in the past has been based on pertinent cost-justification data is not surprising. As the commission noted:
"Historically in this State, and elsewhere, each customer within a defined service classification has paid for his electricity at a rate calculated through an averaging process: the rate he paid was related not to his individual consumption habits and the costs they imposed on society, but to the consumption characteristics of the entire group. This was essentially true even under the most enlightened regulatory practice, which involved trying to tie the rates charged the various groups of customers to the cost of serving them. The costs were the costs of serving the group, not the individual: through the use of one or more cost of service allocation methodologies, aggregate revenue responsibility would be assigned to a service classification in relation to that class contribution to total system costs. Thereafter rates for that class would be established so as, in the aggregate, to achieve the class revenue target.
"In thus designing rates, efforts have been made to allocate greater responsibility to those members of the class more responsible for the costs imposed by the class on the system — for example by the inclusion of seasonal differentials or separate demand charges and rachets — but one principal factor which influences cost has largely been ignored; the time of the day at which service is demanded.”
. It is reported that the charges to some 109 customers in the included class would be less under the new rate schedule than they would have been under the former schedule, with decreases going to 17.7%.
. Under the "two-thirds rule”, the rates charged were to be increased by not more than two thirds of the difference between the old and new rates.
. No particular attention is invited to this third area of cost in our court.