By using chart review to validate electronic alcohol diagnoses, the proposed research will make electronic data more useful in planning and evaluating alcohol treatment, and in conducting research using health plan administrative data. Chart review has been the accepted method for identifying alcohol disorders and other medical problems within health plans. However, as electronic practice databases, notification systems, and medical records proliferate, using these electronic data becomes attractive. Their easy access and low cost make their use inevitable for both research and evaluation, and for resource planning. However, to use them appropriately, we must know-what information will be missed when electronic indices are substituted for indices based on chart review, whether electronic indices will distort inter-group comparisons, and how to adjust prevalence estimates when comparing electronically identified and chart review based prevalence estimates. The proposed analyses use chart review to validate electronic identification of alcohol abuse and dependence in 3 health maintenance organizations (HMOs), of Kaiser Permanente. It is able to make a substantial contribution to this research area because of its large scale - it can examine important characteristics, by which practice might vary. We measure the sensitivity of electronic identification relative to chart review for each HMO, and we also assess heterogeneity in the quality of electronic alcohol diagnoses. We test for gender differences, for lower sensitivity when electronic databases are first initiated, and for differences between outpatient and inpatient diagnoses. We examine how reliably electronic data replicate the chart-review based diagnostic classification of alcohol and other substance abuse problems. Since the optimal observation period for estimating prevalence from health care data is under debate, we compare the sensitivity of one and three-year estimates. The results will enable researchers and health planners to monitor changing service needs effectively, using electronic alcohol diagnoses.