Health Information Systems utilizing computerized medical records abstracting services are widely applied to clinical research, utilization review, health statistics, and similar programs. The utility of such systems is determined primarily by the accuracy of stored abstracts. Even very low per-character error rates can result in a relatively high proportion of abstracts in error. Application of quality control procedures and purposeful design of error identification systems is essential for establishing face validity. Acceptance sampling with comparison of abstracts to medical records can be employed to assure content validity.