An intercropping trial of maize (Zea mays) with two dual-purpose and two forage legumes was undertaken at Dembela Wacho in the Sidamo region of Ethiopia during the 1987 growing season. The objective was to explore the potential of intercropping in order to improve food security, provide high-quality livestock feed and improve soil stability within the existing Borana cropping system. The legumes examined were two annual dual-purpose legumes, cowpeas (Vigna unguiculata) and lablab (Lablab purpureus), and two forage legumes, cassia (Cassia rotundifolia) and centro (Centrosema schottii). The other treatments imposed were four maize planting densities (0, 20, 35 and 50 thousand plants per hectare) and two legume planting times (at maize planting and 20 days later). Cowpeas produced 55% more grain than lablab. Maize planting density did not affect legume grain yield, but monoculture legume grain yield was about twice that of the intercropped ones. Monoculture cowpea grain yield was reduced by 44% when the planting time was delayed by 20 days while the reduction of lablab was only 12%. On the other hand, lablab grain yield was reduced by 56% due to delayed sowing in the intercrop situation. The dry matter yield of cowpeas and lablab was significantly higher than that of cassia and centro. Monoculture legume dry matter yield was significantly higher than the intercropped legume dry matter yield in all cases. The annual combined maize and legume grain and dry matter yields were higher than those in monoculture, thus showing yield advantages for intercropping. The perennial forage legumes did not appear to reduce maize yield when sown at the same time, suggesting that this practice could be a useful strategy in the last year of a cereal cropping cycle to prevent future soil erosion.