Multiple-choice questions (MCQs), the predominant evaluation vehicle in North America, are a reliable and valid measure of knowledge but have many detractors. We use MCQs widely in evaluating undergraduate students in a junior Internal Medicine clerkship but find it difficult to maintain a suitable secure bank of questions. We therefore involved our students in generation of new questions. After one year of this practice we compared the accumulated student-generated questions with those from our faculty. The average facility index and the validity of student-generated questions were comparable to those of the faculty-generated questions. The students' questions were less frequently based on a clinical case scenario than were the faculty-generated questions. We conclude that undergraduate students are an excellent source of MCQs for use in evaluation and that students benefit from involvement in the evaluation process.