||The impacts of maize breeding research for non-temperate environments in the developing world by national agricultural research systems (NARSs) and the International Maize and Wheat Improvement Center (CIMMYT) over 1966-90 are analyzed. Collaborative maize breeding by NARSs and CIMMYT resulted in the development and release of 842 maize varieties and hybrids by public research organizations in developing countries during 1966-90. Of these materials, 53% contained CIMMYT germplasm. Although the share of hybrid releases rose substantially follOWing increased emphasis on hybrids by NARSs, for the period as a whole hybrids comprise just 35% of releases. Maize releases are distributed fairly evenly between tropical lowlands (53%) and tropical midaltitudes, tropical highlands, and the subtropics (47%). In non-temperate ecologies an estimated 24.6 million hectares were planted to improved maize in 1990 (43% of the 57.7 million hectares under maize in the countries under study). Public sector materials accounted for 19.9 million hectares (81%) of the area under improved maize; 4.7 million hectares (19%) were under proprietary materials. Improved maize containing CIMMYT germplasm was planted on 13.5 million hectares in 1990 (55% of the area under improved materials), mostly open-pollinated varieties in tropical lowlands. The fact that more than half (33 million hectares) of the developing world's maize area is still planted to unimproved varieties suggests that significant barriers continue to slow adoption of improved maize germplasm. To reach farmers who do not plant improved maize, breeders must continue developing varieties and hybrids adapted to marginal growing conditions, especially materials tolerant to biotic and abiotic stresses and low levels of external inputs. The still modest use of improved maize germplasm in some regions of the developing world also reflects farmers' difficulty in obtaining seed and points to the need for more effective maize seed industries, especially better seed production and distribution. Finally, to obtain more accurate estimates of the economic impacts of maize breeding research, researchers must monitor 1) the flow of germplasm from public research organizations, especially the use of this germplasm by private sector organizations; 2) farmers' adoption of improved maize; and 3) the yield effects of using improved maize at the farm level.