Cancers of the pleura, lung, and larynx between 1978 and 1989 among active male workers of Electricité de France-Gaz de France were studied in association with asbestos exposure using a case-control design nested within the cohort of workers of the company. The cohort included about 1,400,000 person-years, corresponding to a mean of 117,000 men per year. Exposure to asbestos and to some potential occupational confounders selected among agents from groups I, IIa, and IIb of the International Agency for Research on Cancer was assessed by a job-exposure matrix specific to the company. During the observation period, 12 cases of pleural cancer, 310 cases of lung cancer, and 116 cases of larynx cancer were registered in the cancer register of the company social security department. Four controls per case, matched for year of birth, were randomly selected among the cohort. Conditional logistic regression was used to estimate the odds ratios. A first analysis was conducted in order to assess the validity of the job-exposure matrix by investigating already known relationships between asbestos exposure and asbestosis. For asbestosis, a strong exposure-response relation was found with an odds ratio (OR) of 57.4 [95% confidence interval (CI): 17.0-194.0] in the highest exposure group. There was an elevated risk of pleural cancer (OR, 4.8, CI, 1.2-19.8). For lung cancer, significant ORs of 2.0 (CI, 1.3-3.2) and 1.9 (CI, 1.2-3.0) were found among the two highest cumulative exposure groups; adjustement for confounders slightly decreased the ORs. Squamous cell neoplasm of the lung was associated with asbestos exposure. The association between larynx cancer and asbestos exposure showed a tendency towards a nonsignificant increase in ORs in the highest cumulative exposure categories; this tendency disappeared when adjusting for occupational confounders. This study showed that occupational exposure to asbestos could increase the risk of pleural and lung cancer in a sector in which exposure levels are not considered to be high compared with other industrial settings.