Exposure to variable hyperoxia has recently been shown to be much more effective at producing proliferative retinopathy in the newborn rat than exposure to constant hyperoxia. To incorporate a more clinically relevant oxygen-exposure paradigm in our studies, we have now used a cycle between 50 and 10% oxygen and have compared its effects with those found using new exposures to the previously used 80/40% cycle. Starting at birth and continuing for 14 d, rats were exposed to environments that cycled between 50 and 10% oxygen or 80 and 40% oxygen every 24 h. After exposure, some rats were killed for assessment of retinal vascular development. Others were removed to room air for 4 d before killing and evaluation for the presence of abnormal neovascularization--a clinical consequence believed to be promoted by termination of oxygen therapy. The 50/10% cycle resulted in greater retardation of retinal blood vessel development during oxygen than that found in the 80/40% exposure group. After 4 d postexposure in room air, the incidence of preretinal neovascularization was 97% in the 50/10% rats and 72% in the 80/40% group. Clearly, the overall amount of oxygen the subject receives is less critical than other parameters of its administration in producing proliferative retinopathy. Also, the range of variation (40% in both cases) is not the controlling characteristic. Our results suggest that consistency of oxygen level and avoidance of hypoxic levels should be important concerns in neonatal oxygen therapy.