This study investigated the validity of the CSA activity monitor for assessment of the total amount of physical activity in adolescent athletes. Activity data were compared to data on daily energy expenditure and its derivatives measured by the doubly labeled water method. Seven athletes (speed skaters) with a mean age of 18.2+/-1.1 y were monitored twice (off-season and pre-season) by the activity monitor for eight consecutive days. The primary training during the off-season period was running whereas the pre-season period mainly involved skate training (i.e. inline skating, slideboard training, and skating imitations). Activity counts were significantly correlated to all energy estimates during the off-season period (r=0.93-0.96; P<0.01) whereas not during the pre-season period (r=0.32-0.57). A two-way multivariate analysis of variance showed a significant period effect for activity counts (668+/-163 vs. 548+/-91; P=0.026) whereas not for total daily energy expenditure (15.7+/-2.1 MJ x d(-1) vs. 16.0+/-1.0 MJ x d(-1); P=0.71). The relationship between activity counts and total daily energy expenditure seems to be affected by different training conditions. Therefore these circumstances have to be carefully considered in the interpretation of activity monitor data.