The major format for Continuing Medical Education (CME) in the United States is the large group lecture. To test the hypothesis that the problem based learning (PBL) model for medical student education could be successfully adapted for use in CME with practicing primary care physicians (PCPs), PBL adaptations were developed for experiments with both small (5-10 participants) and large (10-40 participants) groups of learners. Approximately 90% of participants in PBL for CME meetings enjoyed and enthusiastically accepted these education models. Physicians who participated in these programs developed the motivation to change their clinical practice behavior; the frequency of this motivation (30%, 65%, and 85% of participants) was directly linked to the length of the program (1, 2, or 3 hours, respectively). This end-of-program motivation to change practice behavior transferred into actual change as measured both by self-report and by direct measurement of prescription writing. In a completely separate experiment, PBL-based CME programs were provided to PCPs participating in a pediatric asthma disease state management project in a Midwestern health maintenance organization; in the very first year of implementation, the project resulted in a significant decrease (P < 0.03) in utilization of urgent care, with a concomitant 75% decrease in the average cost of urgent care services. Finally, 300 CME educators learned how to teach using the PBL for CME format; outcomes such as acceptance of the model, and end-of-program motivation to change practice behavior in programs given by these teachers (beta results) matched those observed in the initial experiments (alpha results). Taken together, these results suggest that for CME, the large group lecture format can be replaced by the small and large group PBL formats.