Three selection procedures, single-seed-descent, selection of the best 5% in both seed size and seed yield per plant were followed in two divergent crosses of chickpea under four environments in order to study the improvement of seed yield. Analysis of variance showed wide ranges of variability for all characters except days to maturity. Seed size (SSB) was found to be superior to single seed (SSD), followed by seed yield (YB), for developing high yielding lines. Fertility and spacing had no influence on the efficiency of any selection procedure indicating that selection under any environment is equally good.