The mechanisms underlying duodenal ulcer (DU) recurrence after endoscopically confirmed healing are unclear. We sought to examine histologic differences in healing induced by omeprazole and nizatidine. This also entailed assessing interobserver variation in endoscopic diagnosis and the correlation between endoscopic and histomorphologic healing. We treated 31 DU patients for 4 weeks with either omeprazole (20 mg daily a.m.) or nizatidine (300 mg twice daily). The healing rates of both groups showed no significant differences (86.7% versus 81.2%; p = 0.5). Good mucosal repair rates did not differ significantly (38.5% versus 69.2% respectively; p = 0.5). Endoscopists' agreement over scar type was 0.80, with the chance of agreement 0.70 (k = 0.34 +/- -0.08). The correlation between macroscopic and histologic appearance of scars was fair, but fully significant (r = 0.48; p < 0.05). We conclude that the study was too small to detect significant differences in healing patterns between the two drugs. The wide variation in endoscopic diagnosis suggests that mucosal repair is best assessed by histologic examination of biopsy samples.