The authors test a model to predict residential outcome among 201 homeless women aged 50 and over. On two-year follow-up, 47% were successfully domiciled. Of 12 variables examined using logistic regression analysis, only 2 variables, perceived support and number ofcommunity facilities attended were significant predictors of being domiciled on follow-up.Three additional variables – absence of psychosis, a lifetime history of less than one year ofhomelessness, and number of entitlements – attained near-significance. Although residential outcome is predicted by a few individual characteristics, what is most striking is the lack of suitable housing options.