This paper describes and compares two methods used to assign exposure categories to 10,766 petroleum refinery employees included in an epidemiological study. The first scheme grouped individuals into six organization (OR) job groups: Administrative, Maintenance, Operations, Laboratory, General and Other. This scheme used "most common administrative department," as determined by computerized job histories. For the second classification scheme, "most common job title" and "most common plant location" were used to group individuals in four ways (IH codes): 1)job type (administrative, maintenance, operations and unknown); 2) contact with refinery processes; 3) exposure to light aromatics; and 4) exposure to heavy oils. Exposure categories for the latter three were none, occasional, routine and unknown. Comparison of the two schemes showed that OR job groups developed from administrative job histories were sometimes useful in classifying employees according to refinery exposures. While OR job groups were acceptable for employees clearly in managerial, maintenance or operations positions, IH codes provided more precise exposure profiles for these three relatively homogeneous groups. For individuals in laboratory positions and those with vague or unspecified department codes (23% of this cohort), the IH classification codes were necessary to group employees by job and exposure.