Gastric injury resulting from nonsteroidal anti-inflammatory drugs is thought to require direct contact of the drug with the gastric mucosa. An inactive form of a drug (as a prodrug) should protect against mucosal damage. Because sulindac sulfoxide has little effect on prostaglandin synthesis until it is reduced to sulindac sulfide after absorption, we performed a double-blind, crossover endoscopic study in 15 normal subjects to compare the prodrug sulindac sulfoxide (200 mg b.i.d.), the active sulfide metabolite sulindac sulfide (100 mg b.i.d., which yields similar sulfide blood concentrations), a positive control (aspirin, 650 mg q.i.d.), and a negative control (placebo). Each drug was taken for 1 week and gastric mucosa were endoscopically assessed before and after 2, 5, and 7 days of dosing. Aspirin predictably damaged the gastric mucosa, whereas the effects of sulindac sulfoxide and sulindac sulfide could not be distinguished from those of the placebo. We conclude that sulindac sulfoxide as a prodrug is not directly responsible for the reduced severity of gastric mucosal lesions. Both sulindac sulfoxide and sulindac sulfide are poorly soluble in acid gastric contents and the reduced damage may relate to the inability of high concentrations of the drug to enter gastric mucosal cells.