An empirical model of natural gas discovery and production in the United States has been developed to express the over-all effect of the various technologic, geologic and economic forces. Validity of this model is supported by the good fit of historical data, conformance with reasonable boundary conditions and applicability to the closely related crude oil discovery and production history. Forecasts from this model are compared with plausible estimates of gas demand and corresponding supply requirements to the year 2000. In no case are the ultimately recoverable reserves a limiting factor. However, it is clearly shown that the ratio of proved reserves to annual production (R/p ratio) cannot be maintained near present levels without requiring unrealistically high discovery rates to meet even conservative estimates of demand. Only by allowing a decline to about a 10-year R/p ratio by the year 1990, and a 6- to 8-year R/p ratio by the year 2000, could demand be met without a major step-up in discovery rate. These declining reserve levels are expected to require development of economic sources of supplemental gas to stabilize the R/p ratio for natural gas. Depending on the criteria used to define need for supplemental gas, 0.4 to 1 trillion cu ft/year would have to become available as early as 1978, but not later than 1990. Capital requirements for supplemental gas capacity, either as imported LNG or as synthetic pipeline gas from coal, would be on the order of $1 billion per year. Introduction: The rapid increase in demand for natural gas in the U. S. has created some concern for the adequacy of long-rangesupply. Many techniques have been used to forecast demand or supply, but none has successfully related economic and technologic variables. The relationships between such diverse factors as incentives for gas and oil exploration, economics of exploiting proved reserves, geologic limitations on ultimate supply, price regulation, investor confidence and competitive forces controlling the energy market are too complex for individual analysis and systematic correlation. This study of future gas supply is based on a previously developed empirical method for forecasting U.S. natural gas production and discoveries from historical data. In the earlier work, extrapolations of relationships between total (cumulated) production and discoveries of natural gas, and between the life index or working inventory of proved reserves (proved reserves-to-annual production ratio) and time, were used to calculate the future deliverability of natural gas. Various values for ultimately recoverable reserves and lower limits of the life index were used as parameters. Projections based on this earlier work had a high degree of uncertainty, and no attempt was made to establish their relationship with crude oil production and discoveries. In a more recent study, Hubbert proposed a mathematical model that fits both crude oil and natural gas discovery and production. Unfortunately, this model tends to give extremely conservative projections of future supply, probably because of the simple form of the functions imposed on the historical data. In this study. some of the limitations of the previous work have been overcome to achieve a greater degree of certainty in the forecasts. The correlations developed here are not explicit in the independent variable (time), but give a good fit to the historical data for both gas and crude oil data. and also meet plausible boundary conditions. Although empirical, the correlations, in conjunction with Hubbert's work, indicate the possibility of developing fundamental relationships governing the discovery, production and depletion of these natural resources under a reasonably uniform economic and technologic environment.