Introduction of more non-computer tasks has been suggested to increase exposure variation and thus reduce musculoskeletal complaints (MSC) in computer-intensive office work. This study investigated whether muscle activity did, indeed, differ between computer and non-computer activities. Whole-day logs of input device use in 30 office workers were used to identify computer and non-computer activities, using a range of classification thresholds (NCTs). Exposure during these activities was assessed by bilateral electromyography recordings from the upper trapezius and lower arm. Contrasts in muscle activity between computer and noncomputer activities were distinct but small, even at the individualized, optimal NCT. Using an average groupbased NCT resulted in less contrast, also if stratified by subgroups (job function, MSC). Computer activity logs should be used cautiously as proxies of biomechanical exposure. Conventional non-computer tasks may have a limited potential to increase exposure variation in computer-intensive office work.