The construction, validation, and use of a numerical index for measuring the provision of pharmaceutical care are described. The 681 respondents to the 1987 Great Lakes Pharmacy Services Survey were randomly divided into two equal groups of hospitals. Data for the first group (n = 341) were used to construct and validate the pharmaceutical-care index (PCI); data for the second group were used for index analysis. Bivariate analysis of 14 major inpatient pharmaceutical services resulted in one service, admission medication histories, being dropped from the index. Multivariate analysis showed that the remaining services contributed equally to the PCI; they were therefore retained. The possible range of PCI scores was -11.166 to 26.518, with a high score indicating greater provision of service. Analysis of data for the second randomly selected group of hospitals (n = 340) showed that PCI scores differed significantly on the basis of hospital size, hospital teaching affiliation, and pharmacy director's education. The correlation between average daily census and PCI score was fair. Higher scores were associated with the presence of a clinical coordinator or a clinical pharmacist. There were weak associations between PCI score and numbers of pharmacists, pharmacy managers, drug distribution pharmacists, and clinical pharmacists. However, the number of decentralized pharmacists and the number of technicians both showed a fair association with PCI score. Hospitals that had pharmacist participation in ambulatory-care clinics or a staff development program had higher PCI scores than hospitals that did not. The provision of inpatient pharmaceutical services, as assessed by the PCI, may be influenced by hospital size, teaching affiliation, the education of the pharmacy director, and other factors. Further research is needed to extend these findings to other hospitals, expose interactions among the factors that affect pharmaceutical care, and refine the PCI.