1. The mechanism underlying stromal oedema subsequent to epithelial hypoxia was investigated in isolated rabbit corneas. 2. Stromas swelled about 20 micrometer following a 1 hr period of tear side hypoxia in both whole corneal isolates and in preparations in which fluid movement across the endothelium was blocked with silicone oil. In the experiments using whole corneas, stromal thickness was independent of tear side oxygen tension as long as aqueous humour PO2 was greater than 40 mmHg. 3. Neither epithelial thickness nor epithelial electrical resistance, a measure of total ion permeability, was significantly affected by blocking respiration. 4. A 10 degrees C reduction in corneal temperature markedly reduced the rate of hypoxic swelling, suggesting the involvement of a metabolism-dependent hydrating process and implicating the stromal accumulation of a catabolyte. 5. When CN- was used to mimic the hypoxic effect in isolated whole corneas, the passive 36Cl unidirectional flux was unaffected, but lactate production rate and stromal [lactate] more than doubled. 6. These measurements were used with a mathematical model for corneal hydration dynamics to examine the causes of hypoxic oedema. The principal conclusions were: epithelial hypoxia enhances epithelial lactate production and release to the stroma; this process causes an increase in stromal lactate concentration and a decrease in stromal NaCl concentration (primarily through dilution); stromal lactate accumulation exceeds in osmotic load and dilutional effect on [NaCl], producing stromal oedema. Whereas hypoxia produces corneal metabolic acidosis, effects on endothelial permeability of HCO3- transport need not be postulated to explain the stromal oedema that results from hypoxia.