D-Xylose absorption testing is a simple, low-cost method of screening for small intestinal malabsorption. The optimum method to measure D-xylose absorption (serum vs. urine testing) is uncertain. We present a method of improving the accuracy of D-xylose testing. Fifty-one consecutive patients (40 with chronic diarrhea and 5 asymptomatic patients with renal insufficiency) and 6 volunteers with normal renal function were recruited. All received D-xylose, 10 g intravenously and 25 mg orally, on two separate occasions. Serum concentration was determined at baseline and at frequent times thereafter. Area under the curve was calculated to infinity, and D-xylose bioavailability (F) was calculated. A nonlinear model used to derive the relationship between 3-hour D-xylose concentrations and F showed that a value of less than 22.5 mg/dL correlated with an F of less than 60% (malabsorption of D-xylose). A 1-hour D-xylose of less than 20 mg/dL was considered abnormal. Using these indexes for normal 1-and 3-hour D-xylose levels, 90% of patients with D-xylose malabsorption were identified (sensitivity, 90%; specificity, 95%), which represents a marked improvement within the conventional 1-hour D-xylose of less than 20 mg/dL alone (sensitivity, 71%; specificity, 100%). The model was applied prospectively to 15 additional patients with chronic diarrhea. Of these, 12 patients with an F of less than 60% were identified, including 2 patients with normal 1-hour D-xylose levels. Thus, the addition of a 3-hour D-xylose serum level of less than 22.5 mg/dL to conventional 1-hour D-xylose determination greatly improves the D-xylose test for malabsorption screening.