In high productivity zones of the Indo–Gangetic Plain Region (IGPR) of India, the rice–wheat cropping system (RWCS) is exhibiting a decline in factor productivity, low N use-efficiency and deterioration in soil health. Inclusion of legumes in RWCS is advocated as one of the promising agro-techniques to sustain productivity. A field experiment was conducted during 1996/97 and 1997/98 on sandy loam (Typic Ustrochrept) soil of Modipuram, India (29°4′N, 77°46′E, 237 m above sea level) to explore the possibilities of inclusion of cowpea as a break crop in RWCS. The effect of rice or cowpea was measured on the yield, fertilizer N requirement and N use-efficiency in wheat, and also on the soil organic carbon and available N content. The yields of wheat following cowpea were significantly (P<0·05) greater by 19–20%, compared with those following rice. In a cowpea–wheat rotation economic optimum dose of fertilizer N for wheat was smaller than that in a rice–wheat rotation due to increased N availability, better growing environment and greater N use-efficiency (measured as agronomic efficiency, apparent recovery and partial factor productivity) in the former case. On termination of the experiment, soil organic carbon and available N content increased over initial content with increasing rates of fertilizer N under both rotations, but the magnitude of increase was greater in treatments where cowpea preceded wheat.