The authors used an experimental design to compare the effectiveness of unstructured collaborative practice with individual practice on achievement on a complex well-structured problem-solving task. Participants included postsecondary students (N = 257) from a liberal arts college serving primarily nontraditional students and from 2 state universities. Three videotaped instructional procedures were used: lessons on (a) introductory set theory, (b) a problem-solving heuristic, and (c) problem-solving modeling. Participants also engaged in active practice. A posttest assessed participant skills. A 2 (individual vs. collaborative treatment condition) x 2 (nontraditional vs. traditional educational setting) analysis of variance revealed significant main effects for treatment condition. Students who practiced individually outperformed those who practiced collaboratively. Limitations and implications for future research are discussed.