Studies were conducted to determine the optimum number of testers required to rank lines according to their GCA and to understand the quantitative relationship between per se performance and combining abilitiy of the lines in sunflower. Forty-eight lines were crossed to 3 types of testers; an inbred, an open pollinated population and a single cross hybrid. The hybrids were evalu ated during summer 1994. Spearman's rank correlation coefficient indicated that the 3 testers as well as 2 tester combinations ranked the lines similarly for GCA. The correlatlons between per se performance and GCA were not significant for the studied characters except for days to 5O per cent flowering, days to maturity, plant height and oil content. Hence per se performance cannot be taken as a criterion for high GCA for yield and means of its components.