A cross-sectional study was made of the appropriateness of admissions and hospitalization days in an high-technology, acute-care teaching hospital. This 952-bed institution with 31 hospital services is a reference center for the metropolitan area of Barcelona. A systematic random sampling (1:11) of the registers of hospital discharges was carried out during the months of January, February and March 1991 (n = 7,142) excluding discharges from the pediatric, obstetrics and psychiatric departments. A retrospective analysis was made of 639 patients' medical records using the Appropriateness Evaluation Protocol (AEP). The rate of inappropriate hospital admissions was 9.1% (58/639) (95% confidence interval: 6.9-11.5). Inappropriate admissions were primarily attributable to hospitalizations for diagnostic and/or therapeutic services that could have been rendered on an ambulatory basis (70.7% of cases). Overall 29.2% of hospitalization days (1,963/6,731) were inappropriate (95% confidence interval: 28.1-30.3). Inappropriate admissions to hospital, an optimizable health care planning and a conservative physician's attitude (postponed discharge) were the most frequent reasons for unnecessary days of care. The interrater reliability in the use AEP criteria for appropriateness of admission showed a kappa index of 0.5 and that for appropriateness of hospitalization days 0.67. In summary, instruments, such as the AEP protocol, that allow us to determine which patients can be attended at other levels of the health care system and which levels are the most appropriate, are key elements in improving the appropriateness of hospital use in public health care systems.