Several ergonomic studies have estimated computer work duration using registration software. In these studies, an arbitrary pause definition (Pd; the minimal time between two computer events to constitute a pause) is chosen and the resulting duration of computer work is estimated. In order to uncover the relationship between the used pause definition and the computer work duration (PWT), we used registration software to record usage patterns of 571 computer users across almost 60,000 working days. For a large range of Pds (1–120s), we found a shallow, log-linear relationship between PWT and Pds. For keyboard and mouse use, a second-order function fitted the data best. We found that these relationships were dependent on the amount of computer work and subject characteristics. Comparison of exposure duration from studies using different pause definitions should take this into account, since it could lead to misclassification. Software manufacturers and ergonomists assessing computer work duration could use the found relationships for software design and study comparison.