Background Prior educational interventions to increase seeking evidence by medical students have been unsuccessful. Methods We report two quasirandomized controlled trials to increase seeking of medical evidence by third-year medical students. In the first trial (1997–1998), we placed computers in clinical locations and taught their use in a 6-hour course. Based on negative results, we created SUMSearch(TM), an Internet site that automates searching for medical evidence by simultaneous meta-searching of MEDLINE and other sites. In the second trial (1999–2000), we taught SUMSearch's use in a 5½-hour course. Both courses were taught during the medicine clerkship. For each trial, we surveyed the entire third-year class at 6 months, after half of the students had taken the course (intervention group). The students who had not received the intervention were the control group. We measured self-report of search frequency and satisfaction with search quality and speed. Results The proportion of all students who reported searching at least weekly for medical evidence significantly increased from 19% (1997–1998) to 42% (1999–2000). The proportion of all students who were satisfied with their search results increased significantly between study years. However, in neither study year did the interventions increase searching or satisfaction with results. Satisfaction with the speed of searching was 27% in 1999–2000. This did not increase between studies years and was not changed by the interventions. Conclusion None of our interventions affected searching habits. Even with automated searching, students report low satisfaction with search speed. We are concerned that students using current strategies for seeking medical evidence will be less likely to seek and appraise original studies when they enter medical practice and have less time.