Problem: At present, homelessness in the United States is primarily addressed by providing emergency and transitional shelter facilities. These programs do not directly address the causes of homelessness, and residents are exposed to victimization and trauma during stays. We need an alternative that is more humane, as well as more efficient and effective at achieving outcomes. Purpose: This article uses research on homelessness to devise alternative forms of emergency assistance that could reduce the prevalence and/or duration of episodes of homelessness and much of the need for emergency shelter. Methods: We review analyses of shelter utilization patterns to identify subgroups of homeless single adults and families with minor children, and propose alternative program models aimed at the particular situations of each of these subgroups. Results and conclusions: We argue that it would be both more efficient and more humane to reallocate resources currently devoted to shelters. We propose the development of community-based programs that instead would focus on helping those with housing emergencies to remain housed or to quickly return to housing, and be served by mainstream social welfare programs. We advocate providing shelter on a limited basis and reserving transitional housing for individuals recently discharged from institutions. Chronic homelessness should be addressed by permanent supportive housing.