Two experiments were conducted in the same area, in consecutive years, whose objectives were to evaluate silviagricultural systems involving sabiá, jurema-preta, maize and cowpea. In the experiment-1 (taungya), each tree was grown single and intercropped with annual, which were also grown single. We used the design of randomized blocks, with five replications. In Experiment-2 (alleys), the tree were cut at the time of sowing of annual, being young branches and leaves embedded in consortium areas, and used two cultivars of maize and two traditional varieties of cowpea. In the experiment-1, the yield of corn grain was higher in monocultures, with no difference for mass marketable unhusked green ears (MMUGE). Cropping systems did not differ for yields of green beans and cowpea dry. The tree did not differ in their effects on green ear yield and grain, as well as yield of green beans and cowpea dry. The syndications were economically advantageous when the corn was marketed as green ears and cowpea as green beans. In the experiment-2, the tree continued to grow after the consortium, and the jurema grew more than sabiá from 240 days after sowing. The yields of grain and corn were higher in monocultures, with no difference between cultivars. There was no difference among tree crops and cultivars in their effects on grain yield. To MMUGE, there was interaction tree species x rows of positions, and cultivars (30F35H and AG4051PRO) did not differ for this feature. The yield of cowpea grains were higher in monocultures. The variety of cowpea Lagoa de Pedra was superior to José da Penha in the yield of dry beans, with no difference for green beans. There was interaction among traditional varieties x rows of positions to yield dry and green beans, and tree species x positions of rows to yield green beans. Single annual crops provide the highest values of fresh and dry weed. There was no culture systems effect on the chemical characteristics of the soil