ABSTRACT Direct application of low-grade rock phosphate (RP) is a less expensive alternative than the application of chemical phosphorus (P) for small holders relying on nutrient-deficient weathered soils. However, the effect of RP application on growth and yield is unstable for different crop species and soil water conditions. To identify legume species effective for RP direct application, five tropical legumes (cowpea, groundnut, bambara nut, mung bean, and soybean) were tested under two soil types with different water-holding capacities in Sudan Savanna; Lixisols and Plinthosols. The experiment was conducted under three conditions of P treatment, namely, RP application, triple super phosphate (TSP) application, and no phosphorus fertilizer application (0P). Under RP treatment, higher grain yield was obtained in Lixisols, which had a higher water-holding capacity. However, the yield increase was suppressed in the year with more rainfall owing to drastic changes in the soil moisture content, which caused excessive soil moisture stress during vegetative growth periods. Among the five legume species, cowpea and groundnut showed higher grain yields under the RP treatment than did other species. RP application increased shoot biomass but caused small changes in the shoot P and N concentrations. Soybean in Plinthosols had the highest yield ratio of RP/0P, but the absolute yield was low because of the insufficient soil water availability caused by the low water-holding capacity. RP direct application can assist shoot growth and yield, but it rarely compensates for basal yield differences between soil types and among legume species.