Mung bean (Vigna radiata), moth bean (Phaseolus aconitifolius) and clusterbean (Cyamopsis tetragonoloba), grown over three successive years under low and variable rainfall on loamy sand soils of arid western Rajasthan, did not reveal any marked effect of phosphate application (0, 20, 40, 60 and 80 kg P2O5 ha-1) on the consumptive use of moisture. The effects on dry matter production and seed yield were marginal, but not significant. Uniform distribution of precipitation during the growing period, rather than its quantum, had the more favourable influence on plants. P application induced a small increase in the available P status of the soil and also in N and P uptake. But the weight of nodules per plant and root CEC progressively increased with increasing level of P up to 40-60 kg P2O5 ha-1. Phosphate application also lead to an increase in soil N, particularly of the hydrolysable organic- N fraction. Effects on mineralized N were marginal. The amount of N2 fixed was greater in the mung bean and moth bean than in clusterbean. While the phosphate levels did not have any effect on organic- N fractions increased more under clusterbean. While the phosphate levels did not have any effect on the succeeding pearl millet (Pennisetum typhoides) crop, legumes significantly promoted its yield equivalent to > 80kg N ha-1, despite the prevalence of acute drought conditions. The beneficial effect of clusterbean was found to be the greatest, followed by moth bean and mung bean. It seems that the beneficial effect of legume cultivation arouse not only from the total N2 fixed but also from the level of mineralized and hydrolyzable organic N contributed by plant residues left in the soil.