In a clinical trial of 178 patients with pulmonary arterial hypertension, treatment improved the 6-minute walk (6MW) distance but not the peak VO2. To clarify this discrepancy, we examined the exercise data from all study sites. Patients received either the endothelin receptor antagonist sitaxsentan or placebo and underwent serial 6MW and cardiopulmonary exercise testing (CPET). In 518 pairs of body weight-adjusted and unadjusted 6MW and CPET data, the correlation between 6MW and peak VO2 was 0.48. In the 4 sites with the highest overall correlation at baseline (r = 0.62 compared with 0.46 for the other 19 sites, p = 0.04), the correlations at baseline (0.66) and at week 12 (0.65) were similar (p = 0.90). However, the correlation increased significantly from baseline (0.34) to week 12 (0.54, p = 0.0005) for the other 19 sites. The correlations between weight-adjusted 6MW distance and peak VO2 for all centers (0.76) were significantly higher than the unadjusted correlations (0.48, p < 0.0001). Improvement in the correlations over time between the 6MW and CPET data at less-experienced sites was most consistent with improved technical skill with increasing experience. Weight adjustment of the 6MW improved its correlation with peak VO2. In conclusion, in future multicenter trials, CPET expertise should be validated at all sites before subject enrollment.