Objectives-Employees in alumina refineries are known to be exposed to a number of potential respiratory irritants, particularly caustic mist and bauxite and alumina dusts. To examine the prevalence of work related respiratory symptoms acid lung function in alumina refinery employees and relate these to their jobs.Methods-2964 current employees of three alumina refineries in Western Australia were invited to participate in a cross sectional study, and 89% responded. Subjects were given a questionnaire on respiratory symptoms, smoking, and occupations with additional questions on temporal relations between respiratory symptoms and work. Forced expiratory volume in I second (FEV1) and forced vital capacity (FVC) were measured with a rolling seal spirometer. Atopy was assessed with prick skin tests for common allergens. Associations between work and symptoms were assessed with Cox's regression to estimate prevalence ratios, and between work and lung function with linear regression.Results-Work related wheeze, chest tightness, shortness of breath, and rhinitis were reported by 5.0%, 3.5%, 2.5%, and 9.5% of participants respectively. After adjustment for age, smoking, and atopy, most groups of production employees reported a greater prevalence of work related symptoms than did office employees. After adjustment for age, smoking, height, and atopy, subjects reporting work related wheeze, chest tightness, and shortness of breath had significantly lower mean levels of FEV1 (186, 162, and 272 mi respectively) than subjects without these symptoms. Prevalence of most work related symptoms was higher at refinery 2 than at the other two refineries, but subjects at this refinery had an adjusted mean FEV1 >60 mi higher than the others. Significant differences in FVC and FEV1/ FVC ratio, but not FEV1 were found between different process groups.Conclusions-There were significant differences in work related symptoms and lung function between process groups and refineries, but these were mostly not consistent. Undefined selection factors and underlying population differences may account for some of these findings but workplace exposures may also contribute. The differences identified between groups were unlikely to be clinically of note.