A technique for classifying hospitals on a multidimensional basis was developed. Three major sets of attributes were examined: patient case mix, facility mix, and personnel mix. Using multivariate techniques (factor analysis, cluster analysis, and discriminant analysis), 83 variables were examined for 547 hospitals. A total of six factors were identified as useful in grouping comparable hospitals: extended and special services, psychiatric services, radiological services, rehabilitation services, drug dependency services, and pediatric services. The study population included 18 university-owned teaching hospitals and 529 nonuniversity-owned hospitals in 16 western and southeastern states. Through the application of cluster analysis and discriminant analysis, 6 clusters of hospitals were identified, and clusters ranged in size from 10 to 241. University-owned teaching hospitals tended to cluster together in two of the six clusters, providing a basis for comparing to nonuniversity-owned hospitals. The selected hospitals were comparable to Kentucky's two university-owned teaching hospitals: the University of Kentucky and the University of Louisville teaching hospitals. Further analyses will be conducted to examine income and expenditure data of the comparison groups of hospitals in an effort to determine the adequacy of current levels of funding to Kentucky's two teaching hospitals. A list of the 83 study variables and a list of university-owned teaching hospitals and comparable nonuniversity-owned hospitals are appended. (SW)