To evaluate medical insurance claims for chronic disease investigation, claims from eight automotive machining plants (1984 to 1993) were linked with work histories (1967 to 1993), and associations with respiratory, cardiac, and cancer conditions were investigated, in a case-control design analyzed with logistic regression. The primary focus was tool grinding, but other important processes examined were metalworking, welding, forging, heat treat, engine testing, and diverse-skilled trades work. Considerable variability in claim-derived incidence rates across plants was not explained by age or known exposure differences. Asthma incidence increased in tool grinding (at mean cumulative duration: odds ratio [OR], 3.0; 95% confidence interval [CI], 0.90 to 10.0), as did non-ischemic heart disease (cardiomyopathy, cor pulmonale, rheumatic heart disease, or hypertension; OR, 3.1; 95% CI, 1.26 to 7.6). These trends appeared in models with deficits (OR < 1.0) for those ever exposed to tool grinding because of exposure-response miss-specification, demographic confounding, or removal of high-risk workers from the exposed group. The apparent cancer rates identified from claims greatly exceeded the expected rates from a cancer registry, suggesting that diagnostic, rule-out, and surveillance functions were contributing. This study supports the epidemiologic use of medical insurance records in surveillance and, possibly, etiologic investigation and identifies issues requiring special attention or resolution.