The present investigation had three major purposes: (1) to compare the magnitude of risky shifts obtained under conditions of noncompetition and competition; (2) to determine whether the consistent risky-shift effects observed in adults would also be found in young children; and (3) to examine the rationality or irrationality of enhanced risk taking in groups. With risks and payoffs based on monetary gain and loss for problem-solving performance, groups of college students and elementary school children were assigned to competitive, noncompetitive, and control conditions. The major results were as follows: (1) for adults, noncompetitive group decision yielded risky shifts three to five times greater than were obtained in the conditions of competition; (2) the evidence for the risky-shift phenomenon in children was quite weak for both the noncompetitive and competitive conditions; and (3) decision outcomes (problem-solving performance) suggested that the risky-shift effect may be more rational than irrational. On the whole, the results were more consistent with a "responsibility-diffusion" than with a "social value" interpretation of group risk-taking behavior.