The prevalence of nonadherence in IDDM and NIDDM populations and conceptual and methodological issues relevant to measuring diabetes regimen adherence are reviewed. The prevalence of nonadherence varies across the different components of the diabetes regimen, during the course of the disease, and across the patient's life span. Although prevalence rates might be expected to differ between IDDM and NIDDM populations, this rarely has been evaluated. Conceptual problems in defining and measuring adherence include: the absence of explicit adherence standards against which the patient's behavior can be compared; inadvertent noncompliance attributable to patient-provider miscommunication and patient knowledge/skill deficits; the behavioral complexity of the diabetes regimen; and the confounding of compliance with diabetes control. Methods for measuring adherence include: health status indicators, provider ratings, behavioral observations, permanent products, and patient self-reports, including behavior ratings, diaries, and 24-h recall interviews. A measurement method should be selected on the basis of reliability, validity, nonreactivity, sensitivity to the complexity of diabetes regimen behaviors, and measurement independence from the patient's health status. The timing of measurements should be based on the stability of adherence behaviors and temporal congruity with other measures of interest (e.g., indexes of metabolic control). Directions for future research and suggestions for clinical practice are provided.