In a study of eight volunteers and six patients, glibenclamide was placed at different sites of the gastro-intestinal tract under visual control. The dose was instilled once into the stomach and once into the duodenum of the eight volunteers in a randomized crossover design. The six patients underwent diagnostic colonoscopy, and the dose was placed into the ascending colon if pathological findings were not present. The area under the concentration-time curve, completed by extrapolation, and the mean residence time of the drug in the body were calculated. These pharmacokinetic characteristics were examined using a Jonckheere test for ordered alternatives and a Wilcoxon signed rank pair test. The means of the areas under the curve were 477 +/- 131 ng . h ml-1 for the stomach, 475 +/- 142 ng . h ml-1 for the duodenum and 486 +/- 301 ng . h ml-1 for the colon. The mean residence time changed from 2.67 +/- 0.35 h for the stomach to 2.42 +/- 0.48 h for the duodenum and 3.55 +/- 0.68 h for the colon. These results indicate that although glibenclamide is absorbed from all three sites of the gastro-intestinal tract to the same extent, the rates of absorption are different. It is discussed whether these findings really confirm the pH-partition hypothesis in drug absorption. Since glibenclamide--a weak acid--has a pK-value of about 6.5, these data seem to confirm the pH-partition hypothesis of drug absorption.