The purpose of the study is to (a) examine the degree of substitutability between labour and capital, (b) estimate returns to scale, (c) compute the factors' marginal productivities and relative contribution to output and to (d) test the predictive ability of the estimated relationship. These objectives are pursued with the aid of annual time series data for the period 1946 to 1966. Both inter-regional and inter-temporal comparisons have been attempted. The multiple regression technique is applied to various forms of the production functions. It is found that the elasticity of factor substitution is unity in the Indian Sugar Industry. It has experienced increasing returns to scale. Labour as a factor of production is more important both in terms of marginal productivity and contribution to the output and it is more efficient in all-India than in Uttar Pradesh and Bihar and more in Bihar than in Uttar Pradesh. All these findings imply that there is a good scope for employment of more labour and the expansion of sugar industry in India. The paper suggests that the output of a manufacturing industry can reasonably be forecasted through the multiple regression technique.