A data management system and a department-exposure matrix (PAPDEM) was designed and constructed to facilitate exposure assessment for a large multinational study on cancer risks among pulp, paper, and paper product workers. Exposure to 25 major agents was described by prevalence, P (i.e., proportion of the exposed, classified %-range), and level, L (i.e., annual mean concentration at work, classified). Some agents could be assessed only in qualitative terms. The assessment was specific to mill, work department, agent, and time period. The results of industrial hygiene measurements, information from detailed company questionnaires, and the professional judgments of the assessment team were the cornerstones of the assessment. Validity and consistency of the assessment were aimed at by setting default values for P and L prior to the assessment, accurately defining agents and exposure classes, dividing assessment work by subindustry, working in pairs, testing interrater agreement, and finalizing the estimates in a meeting. In spite of these precautions, good agreement between different assessors was difficult to reach. Exposure to chemical agents turned out to be widespread and complex with frequent multiple exposures. A computer-assisted exposure assessment system such as PAPDEM may save time and facilitate assessment in large epidemiological studies requiring complicated exposure assessment procedures. It also provides a good documentation of exposure assignments, which may be useful in the interpretation of the results and in future updates of the study.