The completeness of cohorts identified for historical prospective studies of occupational diseases is subject to many sources of error, including clerical oversight or negligence and poor company record-deeping practices. It is impossible to verify the completeness of occupational cohorts based solely on plant records, since the company files themselves may not be 100% complete. This article presents a method of verification based on IRS records which is completely independent of company-held records. The sensitivity and effectiveness of the method are demonstrated through the application of the verification scheme to an ongoing study of man-made mineral fiber workers in the United States. In six participating plants, 1990 (21.4% of the original cohort) additional records were uncovered following an initial verification and a second complete survey of plant files. It is believed that, without independent verification, errors of this magnitude may be present in many historical prospective studies of this type. An independent verification should be incorporated into all occupational study designs which utilize historical plant records as the primary data base.