Computer-assisted instruction (CAI) using patient simulations was introduced on a pediatric clerkship to teach clinical knowledge and problem-solving skills. To evaluate the efficacy of this educational tool, 151 students were alternately assigned either to complete two CAI exercises or to review the same information in text format. As part of the clerkship examination, all participants completed a latent-image simulation and 10 multiple-choice questions on the specific clinical information presented. After adjusting for prestudy differences in National Board of Medical Examiners Part I scores and cumulative grade point averages, scores on the overall final exam and on the outcome measures were similar for the two groups. In conclusion, CAI simulated patients, when applied in this limited fashion, were as effective but no more effective than text reading in teaching clinical information and problem solving on a pediatric clerkship.