This paper describes a project set up to treat four homeless men using cognitive behavioural therapy (CBT). The referral criteria were that individuals had alcohol and/or substance misuse problems, were roofless (i.e. sleeping rough) immediately before the intervention began and found it difficult or impossible to access hostel places in Southampton. Excessive alcohol use, violence (against self, others and property) and prison sentences were all features of their presentation. The project involved three levels of CBT intervention provided by the clinical psychologist: 1) training for the staff to enable them to work within this model; 2) continued supervision within model to ensure consistency and sustainability; 3) individual formulation (description of the problem within the CBT framework) and psychotherapy. The house itself was also run on a collaborative basis. A number of measures including mental health and social functioning constructs were used to evaluate the project, in addition to some qualitative data. All residents reduced incidents of theft, violence and alcohol consumption. Risk to self and others was also reduced for all residents. Perceived self-efficacy increased slightly for all residents, and staff perceived that they could be more effective, less hopeless, and therefore possibly less stressed as a result of training. More data will be gathered over time.