To determine whether exercise modality affects arterial hypoxemia (EIAH) during training-intensity exercise, 13 triathletes performed 20 min of cycling (C) followed by 20 min of running (R): C-R, and two weeks later, 20 min of R followed by 20 min of C: R-C. Each trial was performed at an intensity slightly above the ventilatory threshold and close to the daily training intensity (75 % of V·O2max). Ventilatory data were collected continuously using an automated breath-by-breath system. Partial pressure of oxygen in arterial blood (PaO2) was measured after each C and R segment and arterial oxyhemoglobin saturation (SpO2) was monitored continuously via pulse oximetry. The metabolic rate was similar across modalities and trials, i.e., C-R (53.8 ± 3.8 vs. 51.1 ± 5.3 ml · min-1 · kg-1) and R-C (52.2 ± 4.5 vs. 53.2 ± 4.6 ml · min-1 · kg-1). EIAH showed significantly greater severity for R compared to C irrespective of the order (p < 0.05 for both trials). R values of PaO2 (and SpO2) for C-R and R-C were 88.7 ± 6.0 mm Hg (93.0 ± 0.6 % SpO2) and 86.6 ± 7.3 mm Hg (93.5 ± 0.6 % SpO2) and C values were 93.7 ± 8.4 mm Hg (95.4 ± 0.4 % SpO2) and 91.4 ± 5.4 mm Hg (94.8 ± 0.3 % SpO2). R ventilatory data described a significantly different breathing pattern than C, with higher respiratory rate (35.9 b · min-1 vs. 51.1 b · min-1 for C-R, p < 0.01; and 50.0 b · min-1 vs. 41.5 b · min-1 for R-C, p < 0.01) and lower tidal volume (2636 ml vs. 2282 ml for C-R, p < 0.02 and 2272 ml vs. 2472 ml for R-C, p < 0.05). We concluded that EIAH was greater during running than cycling for a similar metabolic rate corresponding to training intensity and that EIAH could thus be considered dependent on exercise modality.