A wealth of information exists on ex post returns to research investments for agriculture in the aggregate and for several agricultural commodity groups of commercial importance in the United States. Many of the investments made in agricultural research and, consequently, many of the allocative decisions on research funding, are, however, commodity-specific. Probably the main reason for the limited empirical literature on returns to commodity-specific research is that of the limited production input data available for individual commodities. Data from the U.S. Census of Agriculture, for example, do not permit commodity disaggregation beyond that employed by Bredahl and Peterson. Their analysis which utilizes commodity groups including cash crops, dairy, livestock, and poultry, has been updated by Norton using 1974 Census data. The research reported in this paper represents "early stage" progress in an effort to extend the cross-sectional production function-type analyses of Bredahl, Peterson, and Norton to three specific crops--corn, wheat, and soybeans. The production functions estimated are for the 1977 crop year.