Investigators must often rely on self-reported work history information collected with questionnaires. However, little is known about the agreement between self-reported estimates of exposure and records kept by companies. As part of a cross-sectional medical study of structural fumigation workers, self-reported work history information was collected on both duration and frequency of exposure using an interviewer-administered questionnaire. All company records available on these workers were also collected. Only 15 of 81 structural fumigation companies identified by study participants as current or past structural fumigation employers had records suitable for comparison. These 15 companies employed 32 of the workers who participated in the cross-sectional medical study. The exposure information provided by the 32 workers was compared to information obtained from company records. By examining the agreement between these two data sources, potential limitations were identified in both the self-reported and company-recorded exposure data. By recognizing these limitations in the exposure data, we identified the most appropriate exposure measures to be used in subsequent data analyses. This exercise also demonstrated the difficulties in undertaking these exposure comparisons in an industry consisting of many small, independent companies. Similar difficulties with assessing exposures may be experienced by investigators studying other service industries consisting of many small, independent companies (e.g., dry cleaning, auto repair).