1. In the stomach, prostaglandins protect the gastric mucosa against injuries. One rate-limiting step in prostaglandin synthesis is mediated by prostaglandin endoperoxide synthase (PGHS), the target enzyme of non-steroidal anti-inflammatory drugs (NSAIDs). Two isoforms of PGHS exist: a constitutive (PGHS-1) and an inducible (PGHS-2) enzyme. PGHS-1 is the major source of gastric prostaglandins under physiological conditions. Inhibition of prostaglandin synthesis by traditional NSAIDs such as indomethacin and diclofenac which non-selectively inhibit both PGHS-1 and PGHS-2, causes gastric and intestinal ulceration and delays gastric ulcer healing in chronic models. It has been shown that selective PGHS-2 inhibitors such as L-745,337 (5-methanesulphonamide-6-(2,4-difluorothio-phenyl)-1-inda none) are not ulcerogenic and do not inhibit gastro-intestinal prostaglandin synthesis. However, minimal information is available on the long-term effects of PGHS-2 inhibitors on the healing of previously established gastric injuries. We assessed the cellular localization and expression of PGHS-1 and PGHS-2 during gastric ulcer healing and assessed the effects of L-745,337 on previously established cryoulcers in the rat gastric stomach. 2. PGHS-1 and PGHS-2 were located and quantified by immunohistochemistry during experimental gastric ulcer healing. PGHS-2 immunoreactivity was only negligible in the normal gastric wall, but after gastric ulcerations, it was strongly detected in monocytes, macrophages, fibroblasts and endothelial cells below and between the regenerative glands. PGHS-1 immunoreactivity detected in normal gastric mucosa, disappeared after gastric ulceration in the mucosa adjacent to the ulcer crater. However, it reappeared in the regenerative glands from day 5 onwards. Thus, PGHS-1 and PGHS-2 were located at different sites and their maximal expression followed a different time-sequence. 3. We assessed the effects of L-745,337, indomethacin and diclofenac on gastric ulcer healing and histological healing parameters in rats. L-745,337, indomethacin and diclofenac dose-dependently decreased the healing of gastric ulcers. L-745,337, indomethacin and diclofenac decreased epithelial cell proliferation in the ulcer margin and microvessel density in the ulcer bed on day 8 and increased the thickness of the granulation tissue below the ulcer crater and the gap between both edges of the muscularis mucosae on day 15. Indomethacin and diclofenac, but not L-745,337, decreased synthesis of 6-keto-PGF1alpha and PGE2 in tissue fragments from the stomach and terminal ileum and decreased platelet thromboxane B2 synthesis in clotting whole blood. 4. Dose-response curves for the inhibition of chronic gastric ulcer healing by L-745,337 (administered twice daily intragastrically) showed an ID50 value of 1.7 mg (4.3 micromol) kg(-1). Dose-response curves for the inhibition of PGE2 synthesis in inflammatory exudates in the acute carrageenin sponge rat model, showed ID50 values of 1.1 mg (3.1 micromol) kg(-1) and 1.3 (3.3 micromol) mg kg(-1) for indomethacin and L-745,337, respectively. Thus, inhibition of chronic gastric ulcer healing by L-745,337 occurs within a potentially therapeutic dose-range. 5. In summary, PGHS-2 is markedly accumulated after gastric ulceration in monocytes, macrophages, fibroblasts and endothelial cells in regions of maximal repair activity. Selective inhibition of PGHS-2 by L-745,337 delayed gastric ulcer healing though interference with epithelial cell proliferation, angiogenesis and maturation of granulation tissue in a potentially therapeutic dose range. PGHS-2-derived prostaglandins seem to have an important role in gastric ulcer healing.