This paper presents the results of an intervention strategy designed to decrease drug prescription and increase the use of oral rehydration therapy (ORT) in the treatment of acute diarrhea aimed at family medicine practitioners in two primary health care units of the Mexican Social Security Institute. The intervention consisted of six successive stages: 1) Baseline study of prescribing practices by all 69 physicians in both clinics; 2) Training workshop attended by 36 physicians, including a critical analysis of relevant up-to-date literature, review of results of stage I, discussion of a previously designed treatment algorithm for acute diarrhea, and modification of it according to participant's experience; 3) Post-workshop evaluation; 4) Establishment of a peer review committee to discuss the treatment behavior of participating physicians; 5) Mid-term evaluation for 2 months after the committee stopped functioning; 6) Long-term evaluation at 6, 12 and 18 months, of 20 physicians who received the complete intervention (study group) and 20 physicians who received none (control group). The treatment behaviors of the study and control groups were similar on baseline, but differed significantly (P < 0.01) in the post-workshop evaluation. The study group showed a reduction in the use of antibiotics (from 78.8% to 39.3%) and restrictive diets (47.3% to 12.4%), and increased the use of ORT (31.4% to 58.4%) for children younger than 5 years old with acute diarrhea. In the mid-term evaluation, the use of antibiotics by the study group decreased to 27.6%, prescription of restrictive diets decreased to 6.4%, and use of ORT increased to 73.8% (P < 0.01, in all cases). In the long-term evaluation, persistent positive prescribing behavior was still present in the study group, with a significant difference (P < 0.05) compared to the control group, where no modification was found in the prescribing behavior throughout the study. The average proportion of cases treated according to the algorithm by the study group increased in 29.2% (31.3 to 60.5%) after the workshop, and 45.2% (31.3 to 76.5%) after peer review committee. This behavior was maintained during 18 months after the intervention (74%). The control group showed no significant modification in the average proportion of cases treated according to the algorithm during the study (2 years 7 months). The active participation of physicians in the workshop and in the peer review committee was identified as the key to the short and long-term success of the educational strategy.