Much research work has been conducted in recent years to evaluate the effects of research and extension.2/ Some of these efforts were conducted at the national aggregate level, while others were conducted at the regional or community levels. The methodologies used in those studies varied, but most of the studies were based on regression analysis of ex post data. There are several difficulties associated with the above ex post analyses. First, the time lags from the time of research resource commitment to production of extendable technology (invention), i.e., the leadtime, ranges from a few years to decades. It is difficult to collect time series data to accommodate a long leadtime. Besides, the average lead time is not known unless one studies specific research projects. Thus, most, if not all, researchers ignore the leadtime. They just assume that research output will come about as soon as the research resources are committed, even though most researchers recognize and account for adoption lag. Secondly, research and extension enter into the process of technological innovation at different points in time. Although feedbacks from extension often result in changes in the direction or additional research to be conducted, extension plays little role in the development of a new technology. When the research is completed and a new technology is produced, extension comes into play. Thus, extension follows research in that process. Many researchers recognize this lag, but unless dealing with a specific technology, the lag length is usually unknown. Thirdly, in time series data, research and extension expenditures are highly collinear. One reason for the existence of this collinearity is that the budgets are allocated or authorized for both functions regarding the same issue at the same time. Attempts to estimate the separate effects of research and extension using the regression analysis often fail. Fourthly, all ex post analyses were based on data since World War I when fossil fuel was cheap and abundant. Since the likelihood of having cheap and abundant fossil fuel in the future is very slim, the estimated relationships may not be relevant in predicting or planning for the future. This study proposes an alternative approach to overcome the above difficulties. The purpose of this study is to describe proposed methodology and procedures to estimate the separate rates of return to research and extension in U.S. agriculture using an ex ante analysis.