Two samples of insulin-dependent diabetic youths were studies in successive attempts to refine a brief patient self-report measure of general long-term adherence to a diabetic regimen. Validity and reliability were assessed for both samples by measuring a number of related variables: locus of control, stressful life events, knowledge of diabetes, parent report, and glycosylated hemoglobin. Initial evidence of validity was demonstrated in a variety of criterion areas. Both reliability and validity were replicated on the second sample. Adherence is discussed as a complex multidimensional variable, the measurement of which requires several different methods, including the selfreport measure described here. Warnings about social desirability as a confounding factor were discussed, as were suggestions for other adherence measures.