Objective. The aim of this study was to examine the accuracy of the Calorie Counter personal activity computer (CC method) for athletic activity. Design. An observational study for a week or on a day. Setting. Healthy Japanese volunteers. Participants. Nonathletes (54 males and 53 females) and 38 male athletes, with ages between 13 and 73 years old. Measures. Daily energy expenditure and weekly physical activity vere simultaneously assessed using the CC method and a daily record on physical activity (daily record method) for a week for the nonathletes and one day for the athletes. Results. Mean daily energy expenditures were 9630 (standard deviation, SD, 1250) kJ·d-1 by the CC method, compared with 9530 (SD, 1120) kJ·d-1 by the daily record method for male nonathletes; 7822 (SD, 768) kJ·d-1 and 7790 (SD, 810) kJ·d-1, respectively, for female nonathletes; and 1120 (SD, 1520) kJ·d-1 and 11580 (SD, 1580) kJ·d-1, respectively, for athletes. The correlation coefficients between daily energy expenditures by these two methods ranged from 0.72 to 0.90. Mean daily physical activities were 1520 (SD, 637) kJ·d-1 by the CC method, compared with 1580 (SD, 691) kJ·d-1 by the daily record method for male nonathletas; 1033 (SD, 318) kJ·d-1 and 1080 (SD, 346) kJ·d-1, respectively, for female nonathletes; and 2770 (SD, 914) kJ·d-1 and 3250 (SD, 1090) kJ·d-1, respectively, for athletes. The correlation coefficients between daily physical activities by these two methods ranged from 0.56 to 0.96. Conclusions. Our results suggest that the CC method may be reliable enough for assessing the energy expenditure in athletes as well as in nonathletes during free-living activities.