Used 24-hr recall interviews to assess adherence in a sample of seventy-eight 6- to 19-year-olds with insulin-dependent diabetes mellitus over a 3-month period. Thirteen adherence measures were quantified and grouped into six adherence factors (Injection, Exercise, Diet Type, Testing/Eating Frequency, Calories Consumed, and Concentrated Sweets). Prevailing glucose levels over a 2- to 3-month interval were indexed by glycosylated hemoglobin A1c (HA1c) and glycosylated serum protein (GSP) assays. Fasting triglycerides (TRIG) and total cholesterol (CHOL) assays were used to estimate lipid metabolism. Adolescents were generally less adherent than their young counterparts. Using hierarchical multiple-regression techniques, HA1c and GSP were not reliably predicted by most of the adherence factors; only Calories Consumed showed any predictive power. No significant regression equations emerged for CHOL. In contrast, TRIG was significantly associated with five of the six adherence factors; in all cases, adherence interacted with the patients' metabolic status (as defined by HA1c) at study entry, suggesting that adherence had different effects for youngsters in good versus poor diabetes control.