Background: Scarce attention has been paid to establishing benchmarks for tertiary care for adults with severe mental disorders. Yet, the availability and efficient utilization of residential resources partly determines the capacity of a comprehensive system of care to avoid clogging ever-shrinking acute care bed facilities. Objectives: To describe the actual utilization of and projected needs for residential resources, one part of tertiary care, in the catchment area of a psychiatric hospital in east-end Montreal. To compare results obtained against actual utilization and projected needs evaluated in other Canadian provinces and in other countries, with a view to establishing national benchmarks. Methods: Two surveys were undertaken to establish the number of places in these facilities that were utilized and needed for adults aged 18 to 65 years with severe mental disorders, without a primary diagnosis of mental retardation or organic brain syndrome, and originally from the catchment area. A first survey ascertained the number of places utilized and of those needed for residential care among all long-stay inpatients and all adults in supervised residential facilities. A second survey identified the need for such long-stay hospitalization, nursing homes, and supervised facilities as an alternative or as a complement to hospitalization among acute care inpatients. Results: The actual ratio of places in long-stay hospital units, nursing homes, and supervised residential facilities was 150:100 000 inhabitants. The ideal ratio, according to estimated needs, is 171:100 000. The figure breakdown is as follows: 20:100 000 for long-stay hospital units, 20:100 000 for nursing homes, 40:100 000 for group homes, 40:100 000 for private hostels or foster families, and 51:100 000 for supervised apartments. The needs of this urban, blue-collar population for supervised residential places hovered in the upper range of utilization and standards for European countries and within the proposed standards for Canadian provinces. Discussion: Needs for long-stay hospitalization or for supervised residential facilities cannot be treated as absolute. For example, evaluation conducted in this hospital-led system of psychiatric care may produce higher estimates of institutional care. Comparing actual utilization and projected needs in this urban catchment area with current utilization in other jurisdictions in Canada and Europe should contribute to establishing sound national benchmarks within ranges. Conclusions: It is possible to establish benchmarks that guide the development of supervised residential settings to best meet the needs of the population of adults with severe and persistent mental disorders. The methods used here to assess needs should serve as guidelines for future research, because they were designed to contain the bias of over- or underprovision of care in the current utilization.