Ten men and nine women ages 35 to 55 consumed two diets for 6 weeks each in a cross-over design. The diets were composed of identical natural foods and 30% of the calories as either sucrose or wheat starch. Carbohydrate, fat, and protein supplied 43, 42, and 15% of the calories, respectively. Of the calories 10% was eaten at breakfast (7:00 to 8:30 AM) and 90% at dinner (4:30 to 6:30 PM). Inital body weights were essentially maintained. Fasting serum insulin and glucose levels were significantly higher with the sucrose than with the starch diet. The insulin response and the insulin:glucose ratios after a sucrose load (2 g/kg body weight) were greater after the subjects consumed the sucrose diet. Sucrose feeding produced increases in fasting serum insulin, the insulin:glucose ratio and the insulin response to a sucrose load that were of greater magnitude in a subgroup of nine subjects classified as potentially carbohydrate-sensitive than in normal subjects. Glucose response to a sucrose load and fasting serum glucagon did not differ significantly with diet. Fasting insulin and glucose showed significant increases as a function of time on diet. These results indicate that sucrose feeding produces undersirable changes in several of the parameters associated with glucose tolerance.