This paper describes a morbidity data base for epidemiologic studies which uses information from health insurance claims. Strengths of the data base include completeness and relatively low cost. A limitation is the length of time needed for all claims to be received and processed: rates generated using current information are lower than they will be after all claims are processed. Nevertheless, internal comparisons can presently be made using such rates. Examples illustrate differences in claims experience according to employees' smoking habits and according to kind of job.