Use of bene¢t-cost analysis for economic comparison of agricultural research projects remains confounded, inter alia, by lack of rigour in specifying the withoutproject scenario and how bene¢ts from an innovation endure after its adoption declines. Failure to account for the without-project scenario favours projects to the extent that more bene¢ts are foregone than costs avoided. Moreover, it is unreasonable to assume generally that aggregate bene¢ts from an innovation continue at the peak level until the end of a 30^40 year planning horizon. A general BCA model for agricultural research projects is presented to enable £exible handling of these issues.