The ingestion of non-steroidal anti-inflammatory drugs (NSAID) in 272 patients with bleeding or perforated peptic ulcer was compared with 272 age/sex matched controls. A significantly higher proportion of patients with gastric ulcers had received NSAID than those with duodenal ulcers. Twelve of 90 (13%) patients admitted with bleeding duodenal ulcers had received NSAID compared with 11 of 26 (42%) patients with bleeding gastric ulcers (P = 0.003). Sixteen of 132 (12%) patients with perforated duodenal ulcer were taking NSAID compared with 8 of 24 (33%) patients with perforated gastric ulcer. Thirty eight percent of patients with both bleeding and perforated gastric ulcers had received NSAID compared with 13% bleeding and perforated duodenal ulcers (P less than 0.002). This study confirms the association of NSAID and complicated peptic ulcer in patients of over 65 years and highlights the particular susceptibility of the gastric mucosa to their injurious effect.