This paper analyses the demand for petroleum products in India, For this purpose, econometric models based on time series data are generated for individual products so as to capture product specific factors affecting demand, The models generated follow the non-homothetic translog functional form. The models are validated against historical data by testing them for ex post forecast accuracy, Demand forecasts till the year 2010 are obtained for the various petroleum products using these models, The forecasts indicate a high rate of growth in demand for motor gasoline, high speed diesel oil, kerosene, liquid petroleum gas and aviation turbine fuel, However, the demand for fuel oils, light diesel oil, naphtha and lube oils is expected to grow at a relatively lower rate, To arrive at an optimal refinery process configuration, an analysis of alternative means of satisfying the demand, which include product import options and domestic refining options with alternative process configurations, need to be done, Thus, investment plans for augmenting domestic refining capacity will have to take into account the future pattern of demand, to arrive at an optimal product mix, Forecasting demand forms a crucial aspect in the overall policy analysis of the oil and gas sector.