This paper characterizes the ability of rioprostil, a synthetic primary alcohol prostaglandin E1 analog, to inhibit gastric acid secretion and prevent experimentally induced gastric lesions in rats and dogs, and determines the selectivity (the separation in potency) for these effects. In 4-hr pylorus ligated rats, rioprostil inhibited gastric acid output when administered i.v., s.c., p.o. or intraduodenally, with ED50 values of 0.9, 1.8, 2.9 and 3.7 mg/kg, respectively. Rioprostil suppressed meal-stimulated acid output in Heidenhain pouch dogs and inhibited gastric acid output stimulated by tetragastrin, 2-deoxy-D-glucose, betazole or bethanechol in gastric fistula dogs with ED50 values of 7, 10, 16 and 17 micrograms/kg p.o., respectively. These values in dogs were not significantly different from each other, suggesting that the mechanism of the antisecretory effect of rioprostil is similar regardless of the secretagogue used. Rioprostil prevented ethanol induced gastric lesions in rats (ED50 = 1.5 micrograms/kg p.o.; 12.0 micrograms/kg s.c.) after a 30-min pretreatment. The 8-fold difference in potency between the p.o. and s.c. routes may reflect a local component in the antilesion mechanism of rioprostil. In dogs, rioprostil inhibited aspirin-induced gastric lesions with a p.o. ED50 of 1.6 micrograms/kg. Maximum antilesion activity in dogs for cimetidine or ranitidine was less than 50%, whereas rioprostil inhibited lesion formation by 100% without the appearance of side effects. Oral antilesion selectivity of rioprostil in rats (antisecretory ED50/antilesion ED50) was nearly 2000-fold using the optimum pretreatment time (30 min), and was 12-fold when the pretreatment time (4 hr) was the same as the duration of the antisecretory assay.(ABSTRACT TRUNCATED AT 250 WORDS)