This study examined the effects of hyperoxic training on specific cardiorespiratory and metabolic responses. A group of 19 male subjects trained for 5 weeks on a cycle ergometer at 70 percent of hyperoxic or normoxic maximal heart rate, the hyperoxic group (HG) breathing 70 percent O2, the normoxic group (NG) breathing 21 percent O2. The subjects were tested pre- and post-training under both hyperoxia and normoxia. Measurements included cardiac output (Q(c)), stroke volume (SV), heart rate (HR), pulmonary ventilation (V(E)), oxygen consumption (VO(2)), partial pressure of oxygen (PO(2)), partial pressure of inspired carbon dioxide (PCO(2)), blood lactate concentration [La], and fiber type composition. The V(E) was significantly lower at submaximal work rates (P <0.05) and maximal V(E) increased after training in both groups for both test conditions; hyperoxic V(E) was lower than normoxic V(E) (P <0.05). The maximal V0(2) increased significantly (P <0.05) in both groups for both tests and was 11 percent - 12 percent higher during hyperoxia. Post-training maximal heart rate (HR(max)) was significantly decreased (P <0.05) at the same absolute work rate regardless of the training group or test type. The SV was increased at each work rate and Q c was unchanged. The maximal Q(c) increased significantly (P <0.05) for both groups and types of test: for normoxia: NG 27.3-30.41*min(-1) and HG 30.3-32.31*min(-1) and for hyperoxia: NG 24.7-25.6 and HG 27.9-31.21*min(-1). Although working at the same intensity relative to HR(max), HG showed significantly lower [La] following a single training session, yet maximal values were unchanged after training. Both groups showed a significant increase in the percentage of type IIA fibers post-training but HG retained a larger percentage of IIB fibers. Mitochondrial enzymes; citrate kinase, 3-hydroxyacyl CoA dehydrogenase, and cytochrome c-oxidase were increased in the normoxic trained subjects (P <0.05). In summary, training induced adaptive responses in maximal aerobic power, HR, SV, Q(c), [La], and muscle fiber type composition, independent of inspired PO(2). Intramuscular data suggested there may be some differences between hyperoxic and normoxic training and these were substantiated by mitochondrial enzyme and lactate findings. Our data would suggest that transport mechanisms may limit the ability to increase aerobic power.