The aim of this study was to test the hypothesis that intermittent hypoxic interval training improves sea level cycling performance more than equivalent training in hypoxia or normoxia. Thirty-three well-trained cyclists and triathletes (25.9 +/- 2.7 yr, VO(2max) 66.1 +/- 6.1 mL.min(-1).kg(-1)) were divided into three groups: intermittent hypoxic (IHT, N = 11, P(I)O(2) of 100 mm Hg), intermittent hypoxic interval training (IHIT, N = 11) and normoxia (Nor, N = 11, P(I)O(2) of 160 mm Hg) and completed a 7-wk training program, consisting of two high-intensity (100 or 90% relative peak power output) interval training sessions each week. Each interval training session was performed in a laboratory on the subject's own bicycle, in normoxic or hypoxic conditions for the Nor and the IHT group, respectively. The IHIT group performed warm-up and cool-down plus recovery from each interval in hypoxic conditions. In contrast to IHT, interval exercise bouts were performed in normoxic conditions. Mean power output during a 10-min cycle time trial improved after the first 4 wk of training by 5.2 +/- 3.9, 3.7 +/- 5.9, and 5.0 +/- 3.4% for IHIT, IHT, and Nor, respectively, without significant differences between groups. Moreover, mean power output did not show any significant improvement in the following 3 wk in any group. VO(2max) (L.min(-1)) increased only in IHIT during the training period (8.7 +/- 9.1%; P < 0.05). No changes in cycling efficiency or in hematological variables (P > 0.05) were observed. Four weeks of interval training induced an improvement in endurance performance. However, short-term exposure to hypoxia (approximately 114 min.wk(-1)) did not elicit a greater increase in performance or any hematological modifications.