Strategic priorities are assessed for the agricultural research system in Zimbabwe in a situation characterized by multiple objectives, farm-types, and agro-ecological zones. Economic surplus analysis is used to rank research programs by commodity and research program areas in total and disaggregated by large and small farms in high and low potential regions. No funding, current funding, and 50% more funding are allowed for each program in the analysis. An optimal research portfolio is developed, first with all weight placed on efficiency, and second with increasing weights placed on benefits going to smallholder farmers. Even with no additional weight placed on small-holders, research programs for both small farms and low potential areas enter into the optimal research portfolio. As more emphasis is given to small-holders, the reduction in overall efficiency gained due to research is relatively modest. Maize and cotton were the highest ranked commodity research programs of the 36 commodities considered for both large and small farms. Agronomy and soils research are relatively more important for small-holders, while plant breeding and crop protection are relatively more important for large-scale farmers. © 1999 Elsevier Science B.V. All rights reserved.