The purpose of this study was to compare lactate thresholds (LT) and maximal aerobic capacities (VO(2 max) during sport-specific skating (on ice) and cycle ergometry (off ice) in collegiate hockey players. We hypothesized that VO(2 max) and LT would be higher on ice. We also sought to determine if on-ice and off-ice VO(2 max) values were correlated. Twelve collegiate hockey players performed both graded exercise protocols in randomized order to fatigue. Both protocols included 80 s of work during each stage, followed by 40 s of rest to allow for blood lactate sampling. VO(2 max) was significantly higher on ice (46.9 +/- 1.0 mL*kg(-1)*min(-1)) than off ice (43.6 +/- 0.9 mL*kg(-1)*min(-1); p < 0.05). Maximal heart rate (HR(max)) was also higher on ice (192.2 +/- 1.8 beats*min(-1)) than off ice (186.0 +/- 1.5 beats*min(-1); p < 0.01). LT was drastically higher on ice than off ice as a percentage of VO(2 max) (85.9% +/- 1.9% vs. 69.7% +/- 1.3%; p < 0.01) and HR(max) (90.1% +/- 1.3% vs. 79.4% +/- 1.6%; p < 0.01). Finally, no correlation existed between VO(2 max) values off ice and on ice (r = -0.002; p = 0.99). Our results indicate that off-ice VO(2 max) and LT are not adequate predictors of on-ice VO(2 max) and LT in collegiate hockey players. These findings challenge the use of cycle ergometry to assess aerobic capacity at events such as the National Hockey League Entry Draft combine. We suggest that hockey players be tested in a sport-specific manner, regardless of whether those tests are performed on ice or off ice.