In this study, quantitation of the epithelial proliferation status of gastric glands and duodenal crypts was performed using endoscopic biopsy specimens from patients with rheumatological conditions requiring long-term anti-inflammatory and analgesic therapy, testing the hypothesis that long-term administration of nonsteroidal anti-inflammatory drugs (NSAIDs) may lead to enhanced epithelial cell proliferation in the stomach and duodenum. After a 1-week washout period from NSAID administration, specimens were taken from endoscopically normal gastric antrum and duodenum, and microdissection was used to quantitate mitoses in these sites. Endoscopy and biopsy were then repeated after 2 weeks of NSAID administration, during which patients received in addition either the prostaglandin E1 analogue misoprostol (10 patients) or a placebo (10 patients). Mitotic counts in gastric glands increased similarly and significantly in the two groups from (mean +/- SEM) 4.45 +/- 0.57 to 7.69 +/- 1.08 (P less than 0.001) in the NSAID+placebo-treated group and from 4.31 +/- 0.53 to 7.68 +/- 1.08 (P = 0.006) in the NSAID+misoprostol-treated group. Similar changes took place in the duodenal crypts, from 5.44 +/- 0.68 to 8.60 +/- 1.28 (P = 0.013) in the NSAID+placebo-treated group and from 4.68 +/- 0.56 to 6.35 +/- 0.63 (P = 0.011) in the NSAIDs+misoprostol-treated group. NSAIDs increased the proportion of glands and crypts with bifid or more complex architectural forms. In a small group of patients, misoprostol alone did not alter mitotic rate in glands or crypts over 4 weeks. Thus, NSAIDs increase the rate of proliferation in endoscopically normal gastric and duodenal epithelium of patients with arthritis. This may form one of the mechanisms underlying gastric and duodenal adaptation to NSAIDs.