This paper describes repeated measurements of inhalable flour dust, wheat allergens and alpha-amylase allergens in the bakery industry. A total of 571 full-shift personal dust samples was collected. Wheat allergens and alpha-amylase allergens were measured in 449 and 507 samples, respectively, by the use of recently developed immunoassays. For all three measures of exposure, the main components of exposure variability were determined. Different grouping strategies for studying exposure-response relationships were compared. The specific job of a bakery worker was identified as the most important source of variability in inhalable flour dust concentrations. For exposure to wheat allergens, the job performed was also the most important source of variation, but type of bakery also explained some of the variability. For alpha-amylase allergen exposure, information on type of bakery was more important then job information. For exposure to inhalable dust and wheat allergens, a classification by job title would lead to sufficient contrast in average exposure levels. By contrast, a grouping strategy based on a combination of job and type of bakery appeared to be essential to obtain a useful classification of exposure to alpha-amylase allergens. 1997 British Occupational Hygiene Society.