In the epidemiological approach to occupational cancers, large bodies of data must be analyzed to find rare cases of cancer. The exposure status of workers must therefore be assessed. Inaccuracies will lead to bias toward the null value in certain cases. Job title has often been used as a proxy for exposure status. This study was undertaken to examine content (ie, tasks and activities) associated with job title among men and women in a large Québec municipality. Occupational accident reports were studied for 1589 accidents, and 113 men and women workers were interviewed about job content. Women and men did not seem to have the same accident rates. From interview data, it appeared that women and men with the same job title did not perform the same tasks. Thus, they might have different exposures. The data reported here support caution in using job title to estimate exposure for both genders if the job-exposure matrix has not previously been validated separately by gender. In addition, it may be unwise to adjust relationships between job title and cancer incidence for gender, thus treating gender as a confounder when it may be a proxy for specific exposures.