The objective of this study was to test the hypothesis that high-intensity hypoxic training improves sea-level performances more than equivalent training in normoxia. Sixteen well-trained collegiate and Masters swimmers (10 women, 6 men) completed a 5-wk training program, consisting of three high-intensity training sessions in a flume and supplemental low- or moderate-intensity sessions in a pool each week. Subjects were matched for gender, performance level, and training history, and they were assigned to either hypoxic [Hypo; inspired O2 fraction (Fi(O(2))) = 15.3%, equivalent to a simulated altitude of 2,500 m] or normoxic (Norm; Fi(O(2)) = 20.9%) interval training in a randomized, double-blind, placebo-controlled design. All pool training occurred under Norm conditions. The primary performance measures were 100- and 400-m freestyle time trials. Laboratory outcomes included maximal O(2) uptake (Vo(2 max)), anaerobic capacity (accumulated O(2) deficit), and swimming economy. Significant (P = 0.02 and <0.001 for 100- and 400-m trials, respectively) improvements were found in performance on both the 100- [Norm: -0.7 s (95% confidence limits: +0.2 to -1.7 s), -1.2%; Hypo: -0.8 s (95% confidence limits: -0.1 to -1.5 s), -1.1%] and 400-m freestyle [Norm: -3.6 s (-1.8 to -5.5 s), -1.2%; Hypo: -5.3 s (-2.3 to -8.3 s), -1.7%]. There was no significant difference between groups for either distance (ANOVA interaction, P = 0.91 and 0.36 for 100- and 400-m trials, respectively). Vo(2 max) was improved significantly (Norm: 0.16 +/- 0.23 l/min, 6.4 +/-8.1%; Hypo: 0.11 +/- 0.18 l/min, 4.2 +/- 7.0%). There was no significant difference between groups (P = 0.58). We conclude that 5 wk of high-intensity training in a flume improves sea-level swimming performances and Vo(2 max) in well-trained swimmers, with no additive effect of hypoxic training.