Enterococcusfaecalis is a frequent cause of urinary tract infection in hospitalized patients. Recent reports have suggested that the organism may frequently be acquired by cross-infection from other patients. In this study, we used total DNA restriction patterns to type 135 urine isolates ofE.faecalis from four sets of patients. Isolates were placed into types (all bands identical) and into groups (most bands identical). Most isolates were discriminated by the typing method, and the results suggested that direct cross-infection occurred rarely if at all. However, two groups of clonally related isolates occurred frequently in the urine specimens and also in feces from hospital-associated patients and were often associated with antibiotic resistance. Isolates from these two groups were found less frequently in feces from people not associated with the hospital.Enterococci are among the most frequent causes of hospital-acquired infection, yet our understanding of their spread is very incomplete. Resistance to many antibiotics is already common among enterococci. High-level gentamicin resistance (henceforth referred to as gentamicin resistance) in Enterococcus faecalis has now become widespread. The more recent findings of ,-lactamase production and especially of vancomycin resistance highlight the need to understand routes of infection, in the hope of preventing strains with these properties from becoming endemic in our hospitals.It was long assumed that enterococcal infections were autogenous, arising from organisms already present in the patient's intestinal flora. This view was supported by a study of enterococcal urinary tract infections that used antibiograms for typing and that revealed no clustering of antibiogram types by location in the hospital (6). However, a more recent study of colonization by gentamicin-resistant E. faecalis, typed by plasmid profiles, showed that patients in adjacent beds acquired indistinguishable isolates (26). This finding and other reports of outbreaks (9, 12) have led to speculation that enterococcal infection may frequently be spread between patients, with consequent implications for control methods. The epidemiology of enterococci has been reviewed in detail by Chenoweth and Schaberg (3).The potential for detailed epidemiological studies has been improved recently by the development of DNA fingerprinting methods for typing enterococci. Pulsed-field electrophoresis of restriction fragments has shown the relatedness of a number of P-lactamase-producing E. faecalis isolates across the United States (15), while ribotyping has been used to demonstrate the unrelatedness of vancomycin-resistant E. faecium isolates in one hospital in France (2). In our laboratory, a simpler method that uses conventional electrophoresis of restriction fragments was shown to be highly discriminatory for enterococci (7). In the present study, we used this method to type E. faecalis urine isolates obtained from four sets of patients at the Royal London Hospital over a 9-month period. The findings are corre...