OBJECTIVE To assess the effect of the integration of evidence-based medicine (EBM) in a medical curriculum using small-group discussions with case scenario and problem-based learning on the knowledge, attitudes and skills of medical students. METHODS A quasi-experimental study was conducted in a medical school in southern Thailand during 2008-2009. EBM practice was begun before the completion of the fourth year using case scenario and continued through the fifth year using learner-centred, problem-based self-practice. Knowledge improvement was measured by summative assessment using pre- and post-tests after small-group discussions with a case scenario. Attitudes and skills were measured by self-rating assessments: before initiation of the course (T0) and after the course at week 1, 5, 13, 25 and 37 (T1, T2, T3, T4 and T5), respectively. Data were analyzed using paired t-test and linear mixed-effects model fitted by maximum likelihood. RESULTS One hundred fourteen students took the course, with a mean age of 22.1 years. Before and after knowledge scores showed a significant improvement (4.93 versus 7.43). The proportion of students who achieved the highest knowledge scores was higher (4% at pre-test versus 54% at post-test). Both self-rated attitudes and skills after EBM practice increased dramatically from the initiation of the course (T0; P < 0.001). Compared with T1, significantly lower scores of attitudes and skills were observed at T2 and T3, but both were higher at T4 assessment. CONCLUSIONS Continuous EBM education through small-group discussion and learner-centred, problem-based self-practice can be a useful way to improve a medical student's knowledge, attitudes and skills.