1. Gastric juice was obtained during continuous maximal stimulation with histamine diphosphate, 0.13 μmol h−1 kg−1 (0.04 mg h−1 kg−1) intravenously, in thirty-four control subjects, twenty-eight patients with duodenal ulcer, and eighteen with gastric ulcer. After correction for pyloric losses, the average rate of outputs of water, and of H+, Na+, K+ and Cl− were determined. 2. The results were analysed in terms of the two-component theory of gastric secretion. The acid components in control subjects and duodenal ulcer patients were similar in composition, but that in gastric ulcer patients was by comparison hypotonic and less acid and contained more sodium ions. 3. Calculation of the volume and composition of the alkaline component gave reasonable results in the control and duodenal ulcer groups, but impossible results in the gastric ulcer group. The gastric juice of gastric ulcer patients could not be described in terms of classical two-component theory. 4. In all three groups, volume correlated quite well with Cl− output alone, less well with H+ output, poorly with K+ output, and not at all with Na+ output. However, when volume was expressed as a function of the outputs of Cl− and one other ion, K+ performed poorly, H+ well, and Na+ best of all. 5. Some features suggest that reflux of alkaline liquid from the duodenum is a better explanation of these results than back-diffusion of H+ in exchange for Na+ between the lumen and wall of the stomach. 6. If the refluxing liquid is assumed to be isotonic, its composition can be calculated. Hence, the volume of reflux in a sample of gastric juice can be deduced. Reflux appeared to be more important in the gastric ulcer patients than in the other two groups.