Home visits assessing demented elderly patients yield important information but are perceived to be expensive. We investigated the utility of a home assessment during a clinic visit. A total of 200 demented subjects attending an outpatient geriatric assessment program were enrolled in a prospective, repeated-measures design study comparing the yield of a structured clinic-based home assessment with the yield of a home visit (the criterion standard). A total of 172 subjects completed the protocol. The average age was 76 ± 7 years, 68% were female, 48% married, and the average Mini-Mental Status Examination score was 21 ± 7 points. Only 3 subjects had complete agreement by site, 162 subjects had at least one problem identified only at the home visit, and 7 subjects had at least one problem identified only at the clinic. There were 376 problems identified at both sites, 422 identified only by the home visit, and 478 identified only in the clinic. Problems related to patient safety and caregiver issues were the two most common general categories. Eighty-four percent of the problems identified only at the home visit were potentially serious and 24% of problems identified at both sites were rated as more serious at the home visit. A clinic-based home assessment is not comparable to a home visit for assessing the home environment of demented elderly patients.