This investigation determined the accuracy of self-reports of physical activity compared to observations obtained surreptitiously. Subjects were 44 adults engaged in 1 h of their preferred physical activity while actual activity levels were surreptitiously obtained and compared to immediate self-reported estimates of physical activity. Results indicated that subjects were moderately accurate in recalling their physical activity levels (R = 0.62) but underestimated sedentary activities and overestimated aerobic activities by over 300%. Males overestimated their activity relative to females, and obese subjects underestimated their activity levels compared to normal-weight subjects. Finally, a number of two-way interactions that moderated the accuracy of those subjects engaging in high chronic levels of physical activity were observed.