Expenditures on agricultural research in the public sector, including the International Agricultural Research Centers (IARCs) have stagnated and in some cases, declined sharply in recent years. This has focused attention on issues of efficiency of agricultural research systems, especially the number, size, scope, type, and locations of their programs. This paper examines the issue of research efficiency through a case study of wheat improvement research in developing countries. The basic premise of this study is that the optimal level of research investment should be determined in a global model that incorporates direct research spill-ins. An analytical framework is developed to determine the threshold levels of crop production in a country (or a region within a country) needed to justify crop improvement research programs of different sizes in the presence of spill-ins from abroad. Spill-in coefficients are estimated from yield performance of varieties of different origins grown across a range of environments. The model is then applied to analyze the efficiency of current investments in 69 wheat improvement research programs in 35 developing countries. A major conclusion of the paper is that given the magnitude of potential spill-ins from the international research system, many wheat research programs could significantly increase the efficiency of resource use by reducing the size of their wheat research programs and focusing on the screening of varieties developed elsewhere. ©2000 Elsevier Science B. V. All rights reserved.