Objective: This study examined the effectiveness of collaborative training for individuals with low pretraining self-efficacy versus individuals with high pretraining selfefficacy regarding the acquisition of a complex skill that involved strong cognitive and psychomotor demands. Background: Despite support for collaborative learning from the educational literature and the similarities between collaborative learning and interventions designed to remediate low self-efficacy, no research has addressed how selfefficacy and collaborative learning interact in contexts concerning complex skills and human-machine interactions. Method: One hundred fifty-five young male adults trained either individually or collaboratively with a more experienced partner on a complex computer task that simulated the demands of a dynamic aviation environment. Participants also completed a task-specific measure of self-efficacy before, during, and after training. Results: Collaborative training enhanced skill acquisition significantly more for individuals with low pretraining self-efficacy than for individuals with high pretraining self-efficacy. However, collaborative training did not bring the skill acquisition levels of those persons with low pretraining self-efficacy to the levels found for persons with high pretraining self-efficacy. Moreover, tests of mediation suggested that collaborative training may have enhanced appropriate skill development strategies without actually raising self-efficacy. Conclusion: Although collaborative training can facilitate the skill acquisition process for trainees with low self-efficacy, future research is needed that examines how the negative effects of low pretraining self-efficacy on complex skill acquisition can be more fully remediated. Application: The differential effects of collaborative training as a function of self-efficacy highlight the importance of person analysis and tailoring training to meet differing trainee needs.