Shifts in judgmental extremity were examined when experimental Ss made judgments individually, then as a group, and finally again as individuals. Control Ss made individual judgments twice. A total of 206 secondary school students (103 males and 103 females) comprised the sample. Discussion-to-consensus in experimental groups was followed by significant declines in postconsensus extremity, but under different confidence conditions for males and females. Interpretation of these results stressed the role of group-mediated modification of the cognitive risks and the need for uncertainty reduction reflected in individual extremity judgments.