To assess the impact of a structured, clinically integrated evidence-based undergraduate medicine training programme using a validated tool. DESIGN. Before and after study with no control group. A medical school in Malaysia with an affiliated district clinical training hospital. Seventy-two medical students in their final 6 months of training (senior clerkship) encountered between March and August 2006. Our educational intervention included two plenary lectures at the beginning of the clerkship, small-group bedside question-generating sessions, and a journal club in the paediatric posting. Our primary outcome was evidence-based medicine knowledge, measured using the adapted Fresno test (score range, 0-212) administered before and after the intervention. We evaluated the performance of the whole cohort, as well as the scores of different subgroups that received separate small-group interventions in their paediatric posting. We also measured the correlation between the students' evidence-based medicine test scores and overall academic performances in the senior clerkship. Fifty-five paired scripts were analysed. Evidence-based medicine knowledge improved significantly post-intervention (means: pre-test, 84 [standard deviation, 24]; post-test, 122 [22]; P<0.001). Post-test scores were significantly correlated with overall senior clerkship performance (r=0.329, P=0.014). Lower post-test scores were observed in subgroups that received their small-group training earlier as opposed to later in the clerkship. Clinically integrated undergraduate evidence-based medicine training produced an educationally important improvement in evidence-based medicine knowledge. Student performance in the adapted Fresno test to some extent reflected their overall academic performance in the senior clerkship. Loss of evidence-based medicine knowledge, which might have occurred soon after small-group training, is a concern that warrants future assessment.