There has been little systematic attention to the opinions of faculty in evaluating problem-based learning (PBL). The purpose of this report is to describe the attitudes and opinions of tutors in PBL programs about the relative merits of PBL and traditional medical education, and to examine the influences of selected variables on these attitudes. Questionnaires containing both Likert-type and open-ended questions were sent to 1,287 faculty members who had served as PBL tutors at 22 U.S. and Canadian medical schools during the 1992-93 academic year. All schools with identifiable PBL programs (broadly defined) were included in the survey. The overall response rate was 69% (882 of 1,287). As a group, the respondents were experienced in both PBL and traditional curricula, with an average of 3.75 years of experience in the former and 11.41 years of experience in the latter. Regarding both their overall attitudes and their opinions about seven of nine specific areas, the respondents evaluated PBL more positively than traditional methods. This was especially evident in the ratings of student interest and enthusiasm, faculty interest and enthusiasm, the respondents' personal satisfaction, student reasoning, and preparation for clinical rotations. The two methods were judged to be approximately equally efficient for learning. Traditional methods were judged to be superior for teaching factual knowledge of basic sciences. PBL was particularly popular with faculty in PBL-track programs, with faculty in both the newest and the oldest PBL programs, and with faculty in either primary care or "nontraditional" specialties. The findings confirm and extend the picture of PBL strengths and weaknesses that can be derived from prior anecdotal program descriptions and small sample studies. Experienced faculty seem to prefer PBL in most respects, although they have some serious reservations. The present findings also agree with prior outcome studies of PBL, suggesting that the most strongly held faculty opinions may have a factual basis.