1. We have compared acute gastric bleeding caused by a new slow release preparation of indomethacin (indomethacin Continus) with that caused by aspirin and other indomethacin preparations. 2. In a randomized crossover study, blood loss into timed gastric aspirates was determined in 20 healthy volunteers after receiving, over 96 h, either placebo, aspirin (600 mg four times daily; 17 doses) indomethacin BP (50 mg three times daily; 13 doses), Indocid-R (75 mg twice daily; 9 doses) or indomethacin Continus (75 mg twice daily; 9 doses). A venous blood sample was also taken during each treatment period for subsequent determination of alpha 1-glycoprotein, and for drug assay. 3. Gastric bleeding on placebo was 1.4 (0.7-2.8) microliters 10 min-1 (mean, 95% confidence interval). Both aspirin and the indomethacin preparations caused significantly more bleeding (P less than 0.05). Rates of bleeding after aspirin, indomethacin BP, Indocid-R, and indomethacin Continus were respectively 22.0 (10.7-47.2) microliters 10 min-1, 4.4 (2.2-9.1) microliters 10 min-1, 10.8 (5.3-22.3) microliters 10 min-1, and 5.1 (3.0-10.6) microliters 10 min-1. 4. Rates of bleeding after indomethacin BP and indomethacin Continus, but not Indocid-R, were significantly less than after aspirin (P less than 0.01). 5. Salicylate or indomethacin was detectable in the plasma of all subjects after the active treatment periods, except for one instance involving a subject allocated indomethacin BP. Indomethacin levels were significantly higher 2 h after Indocid-R than with indomethacin BP or indomethacin Continus. 6. alpha 1-acid glycoprotein levels were not significantly affected by prior treatment with aspirin or indomethacin.