This paper argues that many widely referenced studies on the cost effectiveness of alternative assistance programs were conducted at a time when rental housing markets were depressed. Recent increases in rent appear to have reduced the apparent cost advantage that demand-side subsidies hold over supply-side interventions. In addition, the nonsubsidized poor increasingly must compete for a dwindling supply of low-cost privately owned housing. Housing vouchers or similar demand subsidies may be appropriate in some contexts, but economic theory and recent empirical analysis suggest that such subsidies are "not the best at all times and under all situations." Rather, the "best policy" depends on program targeting and the nature and extent of program-induced price increases and externality effects. Since funding limitations currently block the creation of an entitlement housing assistance program, housing policy must balance the often competing goals of expanding the ability of participating low-income households to pay for decent housing while at the same time working to limit the adverse effects that rent increases and the loss of low-cost nonsubsidized stock have on households falling outside of the housing assistance safety net.