Despite the acknowledged importance of agricultural science and technology (AS&T) in enhancing agricultural productivity growth, little is known about the structure of AS&T policy system in developing countries. The structure of this policy system in developing countries was analysed using a “system components-shift effects” framework, with Kenya and Uganda as case studies. The framework was incorporated into the second step technical inefficiency effects translog stochastic frontier model. Food and Agriculture Organisation data for years 1970-2002 were used as input-output data. Important AS&T policy system determinants were identified to include: research expenditures, human capital development, domestic research output, farm-level literacy, intellectual property rights regime, degree of economic openness and access to agricultural extension information.