The objective of this article is to assess the validity of data on occupational activities obtained by a questionnaire. 152 employees from 16 jobs were interviewed independently about activities in their work, using a structured questionnaire. For 11 of the jobs, two further workers were observed during a typical shift, and their activities recorded. The agreement of reporting between subjects doing the same job and with observation was assessed. There was good agreement between subjects and with observation for most of the activities examined when the activities were classified dichotomously. However, frequency of heavy lifting was reported inconsistently. Questionnaires are able to distinguish major differences in physical activity that occur between jobs in the general population, but do not allow more detailed quantification of activities.