To quantify the technical efficiency of diabetes care in family practice settings, characterize the provision of services and health results, and recognize potential sources of variation. We used data envelopment analysis with inputs and outputs for diabetes care from 47 family units within a social security agency in Nuevo Leon. Tobit regression models were also used. Seven units were technically efficient in providing services and nine in achieving health goals. Only two achieved both outcomes. The metropolitan location and the total number of consultations favored efficiency in the provision of services regardless of patient attributes; and the age of the doctor, the efficiency of health results. Performance varied within and among family units; some were efficient at providing services while others at accomplishing health goals. Sources of variation also differed. It is necessary to include both outputs in the study of efficiency of diabetes care in family practice settings.