Housing stability is an important focus in research on homeless people. Although definitions of stable housing differ across studies, the perspective of homeless people themselves is generally not included. Therefore, this study explored the inclusion of satisfaction with the participant's current housing status as part of the definition of housing stability and also examined predictors of housing stability with and without the inclusion of homeless person's perspective. Of the initial cohort consisting of 513 homeless participants who were included at baseline in 2011, 324 (63.2%) were also interviewed at 2.5-year follow-up. To determine independent predictors of housing stability, we fitted multivariate logistic regression models using stepwise backward regression. At 2.5-year follow-up, 222 participants (68.5%) were stably housed and 163 participants (51.1%) were stably housed and satisfied with their housing status. Having been arrested (OR = 0.36, 95% CI: 0.20-0.63), a high level of somatisation (physical manifestations of psychological distress) (OR = 0.52, 95% CI: 0.30-0.91) and having unmet care needs (OR = 0.77, 95% CI: 0.60-0.99) were negative predictors of housing stability. Having been arrested (OR = 0.43, 95% CI: 0.25-0.75), high debts (OR = 0.45, 95% CI: 0.24-0.84) and a high level of somatisation (OR = 0.49, 95% CI: 0.28-0.84) were negative predictors of stable housing when satisfaction with the housing status was included. Because inclusion of a subjective component revealed a subgroup of stably housed but not satisfied participants and changed the significant predictors, this seems a relevant addition to the customary definition of housing stability. Participants with characteristics negatively associated with housing stability should receive more extensive and individually tailored support services to facilitate achievement of housing stability.