Research on the education of diabetic patients in diet management suffers from lack of an adequate method for describing patients' dietary behavior. In this report, a method is proposed for the quantitative assessment of dietary adherence in patients with IDDM. The method relies on comparisons between individualized diet plans and actual consumption as reflected by 24-h diet recalls. Data are presented that suggest this method has reliability and validity. In a sample of 97 patients with IDDM, nearly two-thirds adhered to the number and timing of planned feedings, while only about 10% of patients adhered to planned exchanges, 90% of the time. The average patient added or deleted one exchange for every four exchanges in the diet plan.