Two population improvement methods, mass selection (two cycles) and SI selection (one cycle) were imposed on a base population to study the response to selection by comparing their efficiency in improving autogamy, seed yield and oil content in sunflower. Mass selection was adopted for devel oping two distinct populations regarding maturity, one early (80-85 days) and one medium (95-100 days). The derived populations MSE -2 (early) and MSM 2 (medium) retained substantial variability for most of the characters even after two cycles of mass selection. As compared with the base population, these derived populations showed substantial improvements in seed yield, to an extent of 40 to 65%, oil content (38%) and oil yield (83 to 97%). However, the response to SI selection was effective only in improving oil content. For other characters including seed yield, the derived populations were inferior com pared with the base population.