The Medication Adherence Rating Scale (MARS) is one of the most widely used measurements of adherence in schizophrenia (SZ), but there is no available data regarding its psychometric properties in homeless SZ patients (HSZ). The aim of this study was therefore to assess the psychometric properties of the MARS in a large multicenter sample of HSZ subjects. This multi-centre prospective study was conducted in the following 4 French cities: Lille, Marseille, Paris and Toulouse. Three hundred and fifty-three patients were included. The 3-factor structure of the MARS was confirmed using confirmatory factor analysis: RMSEA = 0.045, CFI = 0.98, TLI = 0.97 and WRMR = 0.76. The unidimensionality of each factor was supported by the satisfactory INFIT statistics. Item internal consistencies were all higher than 0.20 and the Kuder-Richardson were higher than to 0.6, except for factor 2, which was closed to 0.5. Significant associations with symptoms, functioning and quality of life showed satisfactory external validity. The acceptability was satisfactory with missing data lower than 5% for each dimension. The MARS is a short self-administered instrument with acceptable psychometric properties in homeless SZ patients that yields interesting information about medication adherence.