For persons suffering from severe mental disorders (SMD), better quantification of needs may facilitate services planning and evaluation, as well as patient recovery. This study aimed to assess associations of sociodemographic, socioeconomic, service utilization, life events, social support and clinical variables with the level of perceived needs of users with SMD globally and according to their factorial structure. A total of 351 adults with SMD from Montreal (Canada) were interviewed using the Montreal Assessment of Needs Questionnaire and seven other standardized instruments. Clinical records were also consulted. Eight variables were associated with levels of needs. Four factors were extracted from the factor analysis and labeled respectively: Helplessness, Social Integration, Functional Skills and Interpersonal Relationships. Models built from needs assessment factors are stronger than global needs and have a greater association with pertinent variables. Linear regression models based on factors show the importance of satisfaction with life domains and substance abuse in association with higher levels of perceived needs. The recognition of variables associated with each factor would optimize care planning and implementation.