OBJECTIVE: To study the contribution of a normal intake of nutrients to the variability of serum leptin concentrations in persons with type 1 diabetes mellitus. DESIGN: We studied the relation between serum leptin and nutrient intake in a cross-sectional study. METHODS: Serum leptin measured by radioimmunoassay, nutritional data determined by a self-administered 7-day nutritional questionnaire, and the fatty acid composition of the serum phospholipids (measured by thin layer chromatography and gas chromatography) were determined in 60 patients with type 1 diabetes mellitus. Correlation and regression analyses were performed between serum leptin and dietary fatty acids and serum phospholipid fatty acids. RESULTS: In the prediction models for the concentrations of serum leptin in men with type 1 diabetes mellitus, the dietary fatty acids displaced the anthropometric variables, and were independent of the serum testosterone concentrations. This fact remained when the prediction was made on the basis of indirect markers of the intake, such as the serum phospholipid fatty acids. In the women, the fatty acids from the diet or from the serum phospholipids also partly explained the variation in serum leptin, although not displacing the anthropometric variables. CONCLUSIONS: Our data suggest that, in non-experimental conditions, the concentrations of serum leptin in men with type 1 diabetes mellitus and, to a lesser extent, those in women with diabetes, may be influenced by the composition of the habitual diet, especially the type of dietary fat.