It is rare that an architect develops a design theory and subsequent buildingsfrom systematic research. This study set out to test one such design theorydeveloped from research with hostel residents and based on therapeuticmodels (Canter 1979) which was used to design Salvation Army Hostels forsingle homeless People. The study aimed firstly, to discover whether or notOakley's design theory based on therapeutic models, could be incorporatedinto a hostel design; secondly, whether there was any support for the designsand thirdly, whether the staff held similar therapeutic environment.As multivariate places, the hostels were explored by obtaining some meas-ure of resident and staff satisfaction with aspects of the physical environ-ment (Satisfaction Questionnaire), of the therapeutic models in use by staff(Therapeutic Model Questionnaire) and of the conceptions residents have ofthe hostel and similar places (Multiple Sorting Task) . Visits were made tofour Salvation Army Hostels in Bolton, Darlington, Swindon and Liverpooland a total of 54 residents and 45 staff participated in the study.Results indicate that the four hostels are recognised by staff to be homely,bright and colorful places which encourage individual growth, that is, as-pects of the design theory. In general there were high levels of satisfactionwith different aspects of the hostels for staff and residents offering supportfor the existence of different environmental roles. This variation betweenpeople was found further to vary across settings. Both resident and staffsatisfaction with different aspects of the hostels was found to be highest forthe newest, smallest and most sophisticated hostel in terms of design andlowest at the oldest, largest and least and sophisticated hostel. No distinct therapeutic models were identified for staff. Results suggest thehostels are not operating as the designer envisaged, presenting some supportfor the arguement that actual use of a therapeutic facility or indeed anyphysical setting can confound an original design brief (Rivlin and Wolfe1979).The study has implications for the use of therapeutic models as goals for de-sign and has further implications for the role of design theory in the successof any therapeutic environment.