Silvopastoral systems, win-win technologies to enhance productivity and providing environmental services, may play an important role in retaining trees (Ibrahim et al. 1999, Gobbi and Ibrahim 2004), and to mitigate the deforestation process attributed to cattle production in the last decades. Brachiaria grasses are the most widely planted tropical forage grasses in Latin America. It is calculated that the area under Brachiaria grasses in Brazil ranges between 30 and 70 million hectares (Miles et al. 1998). Tree-grass associations with Brachiaria species are becoming frequent in the dry tropics; however, there is little information about how compatible these grasses are in silvopastoral mixtures with native timber tree species. Understanding the ecological and biophysical interactions in these systems is the key for designing and managing silvopastoral systems for sustainable production (meat, milk and timber) and generation of environmental services. Three native timber tree species (Pithecellobium saman, Diphysa robinioides and Dalbergia retusa) were planted with an improved grass (Brachiaria brizantha) and a naturalised grassland dominated by Hyparrhenia rufa. The central aim of this study was to understand the main tree-grass interactions in silvopastoral systems in the dry tropics. Tree growth, grass production, fine root interactions, water and radiation competition were studied to evaluate the most important biophysical interactions in dry tropics. D. robinioides was the fastest growing species, and P. saman the slowest; while B. brizantha produced three times the aboveground and twice the belowground biomass as H rufa. Trees had no effect upon grass yield. P. saman grew more slowly in B. brizantha than in H rufa treatments, but there was no effect of pasture species upon the other two tree species. The carbon in phytomass varied between 7 and 13 Mg C ha·1 in no-tree pastures and silvopastoral systems, respectively, and soil organic carbon (SOC) averaged 100 Mg ha·1 for the top 60 cm of soil. B. brizantha appeared to stimulate tree root production, which in turn was highly correlated with the light fraction of soil organic carbon, resulting in increments in SOC of up to 9.9 Mg ha-1 year-1• Tree fine roots grew more than grass, responding more rapidly to increase in water availability, increasing the possibility of their establishment in grasslands. The fact that trees tended to have higher root competitive capacity in soil deeper layers and under canopy positions, and grass was more competitive in intermediate positions (away from trees) and superficial layers demonstrated the niche differentiation of this association. The introduction of improved and drought tolerant grass species such as B. brizantha enhances the root development of whole systems, increasing the coexistence of components in region with seasonal water deficits. The increment of soil nitrogen content by tree component, attributed to their nitrogen fixing capacity, is an advantage of these silvopastoral systems. The transmission through the canopy of photosynthetically active radiation varied between 46.6 and 68.0% for D. robinioides and P. saman, respectively. Grass species did not affect the sap flow velocity of D. retusa and D. robinioides; whereas P. saman presented a higher sap flow velocity associated with B. brizantha than with H rufa. Tree species did not affect the water use of grasses. At a plot level, these young trees used less water than grasses ( on average, 0.11 vs 1.2 mm dal, respectively), showing the importance of selection of compatible grass species in these dry conditions. D. robinioides and D. retusa proved to be adapted species to seasonal drought via their adaptation in regulating water use according to soil and climatic conditions. The silvopastoral systems evaluated showed great productive and environmental benefits such as the promotion of tree growth through the establishment of drought tolerant and productive grass species; the dry matter production of grass species was not affected by trees after four years; a relative high carbon sequestration in biomass and soil; nitrogen fixing of tree timber species; niche differentiation in terms of soil exploration; and the adaptation of evaluated species to seasonally dry environments.