As part of the evaluation of an educational intervention carried out at primary health care units in Mexico City a home visit was included for patients treated for acute infectious diarrhea. In such visit, effected for 401 patients before the educational intervention and for 406 after the same, it was possible to evaluate compliance with the treatment measured through the amount of prescribed drugs distributed by the institution but not consumed by patients. A significant improvement was observed in compliance with the treatment even though this was not the explicit objective of the intervention. Compliance is related to variables of different dimensions--characteristics of the physician, characteristics of the patient, physician-patient rapport, health condition and health concept--and the educational intervention tends to homogenize the patient's behavior as a result of more emphatic improvement in those subgroups with worst compliance levels.