The validity of retrospective assessment of occupational exposure greatly depends on the amount of detail in the available information, on the knowledge of the specific industrial process by the experts, and on the criteria adopted to define relevant exposure. These criteria are difficult to standardize and are rarely made explicit in published reports, which makes it difficult to interpret inconsistencies among different studies. In two ongoing case-control studies on kidney cancer and, respectively, malignant lymphomas, a detailed occupational history was obtained and supplemented by 19 additional questionnaires, specifically addressing industrial activities where the knowledge of job title alone would have been insufficient for reliable exposure assessment. One further questionnaire was used to collect details of task and environment for all the other activities. These data are used to establish probability, intensity and frequency of exposure to 30 substances known or suspected to be carcinogenic from previous studies. There are two basic steps in the exposure assessment procedure: firstly, general rules are defined for each job within each activity covered by specific questionnaires; secondly the judgement is modulated according to the detailed tasks, working conditions and environment. To illustrate the process and to facilitate comparison with other studies, examples are given for a few common exposures in the textile and metal industries--the two most frequent economic activities in the study area--namely exposure to mineral oils, formaldehyde, aromatic solvents, chlorinated solvents and other organic solvents.