Abstract Legumes can be an important source of N for cereals in tropical, subsistence farming systems that use little or no fertilizers. Our objective was to identify measures of soil N availability, following growth of diverse legumes, that correlated with yield of a subsequent unfertilized maize ( Zea mays L.) crop. Unfertilized maize followed nine 3‐yr‐old systems (six monocultures of planted trees legumes, a groundnut [ Arachis hypogea L.]‐maize‐soybean [ Glycine max (L.) Merr.] rotation, an uncultivated fallow with natural regrowth of vegetation, and unfertilized maize monoculture) in two experiments on a Ustic Rhodustalf in eastern Zambia. Total soil C and N before maize planting were not related to grain yield of maize. Preseason soil inorganic N (NO 3 + NH 4 ), aerobic N mineralization, and light‐fraction N (N in macroorganic matter, 150–2000 µm and <1.37 Mg m ‐3 ) at 0‐ to 15‐ cm depth correlated ( P < 0.01) with maize grain yield. Preseason inorganic N combined with light‐fraction N accounted for 59% of the variance in maize grain yield. The predictability of maize yield was slightly improved (64% of the variance) by including the population of the parasitic weed striga [ Striga asiatica (L.) Kuntze] with preseason inorganic N and light‐fraction N. The results suggest that yield of unfertilized, rain‐fed maize following different rotational systems on a N‐limiting soil in an area with monomodal rainfall was strongly related to the additive effects of preseason soil inorganic N and a soil N fraction related to N mineralization.