Pharmacists' clinical interventions in a group of British hospitals were counted, and a model to determine factors that affected the intervention rate was developed. All pharmacists who visited patient wards in 27 acute care hospitals recorded their daily ward visits and their clinical interventions during five consecutive days (Monday through Friday) in June 1993. An intervention was defined as any recommendation made with the intent of changing drug treatment. Mixed-model Poisson regression was used to try to explain variations in the intervention rate, defined as the number of physician-accepted interventions divided by the number of occupied-bed days. Possible predictors of intervention rate considered were characteristics of the hospitals, the wards, and the pharmacists. During the study period, 248 pharmacists visited 10,478 beds and proposed 3,501 interventions. Of these interventions, 3371 were accepted, 56 were rejected, and 74 were unresolved. The most frequent reasons for the interventions involved the dose (29%), the need for therapy (21%), the choice of drug (14%), and the route (12%). Ward type, pharmacist grade, and the total time the pharmacist spent on the wards were significant predictors of the intervention rate. To validate the model, data were collected during the same period in 1994; the model predicted the number of interventions within 1 of the actual number in 82% of cases. In a model explaining the factors that affected the rate of physician-accepted pharmacist interventions in acute care hospitals in the United Kingdom, ward type, pharmacist grade, and total time spent on the ward by the pharmacist were significant predictors of the intervention rate.