The literature on homeless adults with severe mental illness is generally silent on a critical issue surrounding service delivery--the contrast between housing first and treatment first program philosophies. This study draws on data from a longitudinal experiment contrasting a housing first program (which offers immediate permanent housing without requiring treatment compliance or abstinence) and treatment first (standard care) programs for 225 adults who were homeless with mental illness in New York City. After 48 months, results showed no significant group differences in alcohol and drug use. Treatment first participants were significantly more likely to use treatment services. These findings, in combination with previous reports of much higher rates of housing stability in the housing first group, show that "dual diagnosed" adults can remain stably housed without increasing their substance use. Thus, housing first programs favoring immediate housing and consumer choice deserve consideration as a viable alternative to standard care. (Contains 1 table and 4 figures.)