We hypothesized that tissue hyperoxia would enhance and hypoxia inhibit neovascularization in a wound model. Therefore, we used female Swiss-Webster mice to examine the influence of differential oxygen treatment on angiogenesis. One milliliter plugs of Matrigel, a mixture of matrix proteins that supports but does not itself elicit angiogenesis, were injected subcutaneously into the mice. Matrigel was used without additive or with added vascular endothelial growth factor (VEGF) or anti-VEGF antibody. Animals were maintained in hypoxic, normoxic, or one of four hyperoxic environments: hypoxia -- 13 percent oxygen at 1 atmosphere absolute (ATA); normoxia -- 21 percent oxygen at 1 ATA; hyperoxia -- (groups a-d) 100 percent oxygen for 90 minutes twice daily at the following pressures: Group a, 1 ATA; Group b, 2 ATA; Group c, 2.5 ATA; Group d, 3.0 ATA. Subcutaneous oxygen tension was measured in all groups. The Matrigel was removed 7 days after implantation. Sections were graded microscopically for the extent of neovascularization. Angiogenesis was significantly greater in all hyperoxic groups and significantly less in the hypoxic group compared with room air-exposed controls. Anti-VEGF antibody abrogated the angiogenic effect of both VEGF and increased oxygen tension. We conclude that angiogenesis is proportional to ambient pO(2) over a wide range. This confirms the clinical impression that angiogenesis requires oxygen. Intermittent oxygen exposure can satisfy the need for oxygen in ischemic tissue.