Objectives: Patterns and predictors of engagement in peer support services were examined among 50 previously homeless veterans with co-occurring mental health conditions and substance use histories receiving services from the Veterans Health Administration supported housing program. Method: Veteran peer specialists were trained to deliver sessions focusing on mental health and substance use recovery to veterans for an intended 1-hr weekly contact over 9 months. Trajectories of peer engagement over the study’s duration are summarized. A mixed-effects log-linear model of the rate of peer engagement is tested with three sets of covariates representing characteristics of the veterans. These sets were demographics, mental health and substance use status, and indicators of community participation and support. Results: Data indicate that veterans engaged with peers about once per month rather than the intended once per week. However, frequency of contacts varied greatly. The best predictor of engagement was time, with most contacts occurring within the first 6 months. No other veteran characteristic was a statistically significant predictor of engagement. Older veterans tended to have higher rates of engagement with peer supporters. Conclusions and Implications for Practice: Planners of peer support services could consider yardsticks of monthly services up to 6 months. Peer support services need a flexible strategy with varying levels of intensity according to need. Peer support services will need to be tailored to better engage younger veterans. Future research should consider other sources of variation in engagement with peer support such as characteristics of the peer supporters and service content and setting.