The validity of a diary that estimates exposure to tasks, activities, and postures of the trunk was determined by comparing these self-reported exposure data with observational data of a whole working day. Two populations were studied: 32 professional drivers and five nurses. The nurses and 16 drivers also filled out a shortened version of the diary during another working day. Both versions of the diary showed poor agreement with observations over the same period. However, for variables concerning activities and postures of the trunk agreement was improved by the shortening of the diary. It is concluded that in epidemiologic studies observational measurements of exposure cannot validly be replaced by diaries or similar self-reported exposure data, because the self-reports easily lead to misclassification.