OBJECTIVE: To compare the epidemiology of vancomycin-resistant Enterococcus faecium (VRE) in a long-term-care unit and an acute-care hospital. DESIGN: Point-prevalence surveys for VRE rectal colonization of patients were carried out over a 21-month period in patients in a long-term-care unit and an acute-care hospital (medical ward and intensive-care units). The environment and hands of healthcare workers also were sampled for VRE. Contour-clamped homogeneous electric field (CHEF) electrophoresis was used to evaluate possible transmission among roommates and the relatedness of patient strains to those in the environment and on the hands of healthcare workers. SETTING: A 200-bed Veterans Affairs Medical Center with an attached 90-bed long-term-care unit. RESULTS: From December 1994 to January 1996, rectal VRE colonization of patients in the long-term-care unit increased significantly from 9% to 22%. In contrast, patients on the medical ward rarely were colonized after the first survey in December 1994, and only two intensive-care-unit patients were found to be colonized during the four surveys. The environment was contaminated persistently in the long-term-care unit. In the four surveys, carriage of VRE on hands of healthcare workers varied from 13% to 41%; 65% of healthcare workers with VRE found on their hands worked in the long-term-care unit. Seven different strains were identified by CHEF typing. Although the initial survey found only vanA strains, subsequent surveys showed vanB strains also were present. CONCLUSIONS: Residents of a long-term-care unit frequently were colonized with VRE, but infections were uncommon in this population. The environment of the long-term-care unit was contaminated with VRE, and VRE was found frequently on the hands of healthcare workers in this unit. Both vanA and vanB genotypes were found in this setting.