Serum pepsinogen has been suggested as a risk marker for the development of gastrointestinal side effects of non-steroidal anti-inflammatory drugs (NSAIDs). The relation between serum levels of pepsinogen A (PGA) and pepsinogen C (PGC) and endoscopic findings in association with short-term NSAID use was investigated in two double-blind, crossover studies in healthy volunteers. Thirty-two male subjects with a median age of 23 years were given naproxen, oxindanac, or piroxicam for 2 weeks. Upper endoscopy was performed by the same investigator before and after each treatment period, scoring mucosal injection and erosive and hemorrhagic lesions separately on 150-mm visual analogue scales. Blood samples for pepsinogen analyses were drawn before each endoscopy. PGA and PGC were analyzed by means of solid-phase radioimmunoassays. Significant amounts of gastroduodenal mucosal lesions were found in all treatment periods, whereas neither PGA nor PGC changed during treatment. Moreover, the initial levels of serum pepsinogen also failed to predict the subsequent development of gastroduodenal lesions. The risk of a type-II error was small in this study, and our results therefore do not support the use of PGA or PGC as a risk marker in this context.