A case-control study of occupational risk factors for sinonasal cancer was conducted in France in 1986-1988. The study included 207 histologically confirmed cases and 409 controls. Among the male cases were 59 men with squamous cell carcinoma and 82 with adenocarcinoma. The risk of sinonasal cancer in relation to wood dust exposure was studied in these two groups. The analysis was based on a case-by-case assessment of exposure by an industrial hygienist. Hardwood and softwood were distinguished. An approximate twofold increase in risk for squamous cell carcinomas was observed for cases whose first exposure to either hardwood or softwood occurred before 1945; however, the two types of exposure were highly correlated. An exposure to wood dust--from either hardwood alone or hardwood and other kinds of wood--was found for all but two of the 82 male cases with adenocarcinoma. The effects of different elements of exposure to hardwood (duration, level, period) were studied in detail with a logistic model. Two components of exposure--duration and average level--contributed independently to the overall very elevated risk. Additional exposure to wood other than hardwood did not increase the risk.