In a test of a model of the relationship of ability to group problem solving on a complementary task, 1008 college students were administered the Terman Concept Mastery Test. After being trichotomized as High (H), medium (M), or low (L) ability, they retook the test working with three partners in one of the 15 possible ability combinations of tetrads HHHH, HHHM,…, LLLL. Control H, M, and L individuals retook the test alone. The predicted order of improvement scores from the first to the second test for the three ability levels was largely supported, as was the predicted order of absolute second-test performance for the 15 groups. It is proposed that, on a difficult complementary task, group problem solving performance increasingly exceeds the performance of the same individuals working, independently as the ability level of the individuals comprising the group increases. The failures of prediction of the model are interpreted as indicating that low and/or medium-ability members may hinder the performance of high-ability members of a problem-solving group.