Conceptual and methodological problems plague efforts to prevent homelessness. Attempts to identify individuals at risk are inefficient, targeting many people who will not become homeless for each person who will. Such interventions may do useful things for needy people, but evidence that they prevent homelessness is scant. Subsidized housing, with or without supportive services, has ended homelessness for families and played a key role in ending it for people with serious mental illnesses. Other risk factors may be less important once housing is secured. But programs that allocate scarce housing may simply reallocate homelessness, determining who goes to the head of the line for housing, not shortening the line itself. We recommend reorienting homelessness prevention from work with identified at-risk persons to efforts to increase the supply of affordable housing and sustainable sources of livelihood nationwide or in targeted communities. (abstract from the document)