Adverse drug reactions (ADRs) were examined in 1,960,059 hospitalized Medicare patients in 584 United States hospitals in 1998. A database was constructed from the MedPAR database and the National Clinical Pharmacy Services survey. The 584 hospitals were selected because they provided specific information on 14 clinical pharmacy services and on pharmacy staffing; they also had functional ADR reporting systems. The study population consisted of 35,193 Medicare patients who experienced an ADR (rate of 1.8%). Of the 14 clinical pharmacy services, 12 were associated with reduced ADR rates. The most significant reductions occurred in hospitals offering pharmacist-provided admission drug histories (odds ratio [OR] 1.864, 95% confidence interval [CI] 1.765-1.968), drug protocol management (OR 1.365, 95% CI 1.335-1.395), and ADR management (OR 1.360, 95% CI 1.328-1.392). Multivariate analysis, performed to further evaluate these findings, showed that nine variables were associated with ADR rate: pharmacist-provided in-service education (slope -0.469, p=0.018), drug information (slope -0.488, p=0.005), ADR management (slope -0.424, p=0.021), drug protocol management (slope -0.732, p=0.002), participation on the total parenteral nutrition team (slope 0.384, p=0.04), participation on the cardiopulmonary resuscitation team (slope -0.506, p=0.008), medical round participation (slope -0.422, p=0.037), admission drug histories (slope -0.712, p=0.008), and increased clinical pharmacist staffing (slope -4.345, p=0.009). As clinical pharmacist staffing increased from the 20th to the 100th percentile (from 0.93+/-0.77/100 to 5.16+/-4.11/100 occupied beds), ADRs decreased by 47.88%. In hospitals without pharmacist-provided ADR management, the following increases were noted: mean number of ADRs/100 admissions by 34.90% (OR 1.360, 95% CI 1.328-1.392), length of stay 13.64% (Mann-Whitney U test [U]=11047367, p=0.017), death rate 53.64% (OR 1.574, 95% CI 1.423-1.731), total Medicare charges 6.88% (U=111298871, p=0.018), and drug charges 8.16% (U=108979074, p<0.001). Patients in hospitals without pharmacist-provided ADR management had an excess of 4266 ADRs, 443 deaths, 85,554 patient-days, $11,745,342 in total Medicare charges, and $1,857,744 in drug charges. The implications of these findings are significant for our health care system, especially considering that the study population represented 15.55% of 12,261,737 Medicare patients and 5.71% of the 34,345,436 patients admitted to all U.S. hospitals.