OBJECTIVE - To assess the metabolic effects of chronic dietary fructose consumption in diabetic subjects. RESEARCH DESIGN AND METHODS - Six type I and 12 type II diabetic subjects consumed, in random order, two isocaloric study diets for 28 days. In one diet, 20% of energy was derived from fructose. In the other diet, <3% of energy came from fructose, and carbohydrate energy was derived primarily from starch. Both study diets were composed of common foods. All meals were prepared in a metabolic kitchen where all foods were weighed during meal preparation. RESULTS - Mean plasma glucose, urine glucose, and serum glycosylated albumin values were lower during the fructose diet than during the starch diet, but the differences achieved only marginal statistical significance. The day-28 value for mean plasma glucose was 12.5% lower (P = 0.03) during the fructose diet than during the starch diet. At days 14, 21, and 28, fasting serum cholesterol and LDL cholesterol were both significantly higher during the fructose diet than during the starch diet. The day-28 values for serum cholesterol and LDL cholesterol during the fructose diet were 6.9% (P = 0.008) and 10.9% (P = 0.002) higher, respectively, than the corresponding values during the starch diet. No differences were observed between the study diets in fasting serum HDL cholesterol, fasting serum triglycerides, peak postprandial serum triglycerides, or fasting serum lactate. Peak postprandial serum lactate was significantly higher during the fructose diet. Type 1 and type II diabetic subjects responded to the diets in a consistent way, but type I subjects experienced significantly more hypoglycemia during the fructose diet than during the starch diet. CONCLUSIONS - A high-fructose diet may result in reduced glycemia in diabetic subjects but at the expense of increased fasting serum total and LDL cholesterol.