PURPOSE: To evaluate the mental health status of children working on the streets in Sao Paulo City, Brazil, two years after their participation in a psychosocial program, and to identify factors associated with their mental health status. METHODS: From a total sample of 126 children working on the streets, 107 (85%) were re-evaluated two years after the initiation of a psychosocial program which aimed to cease their work on the streets. The focus was the presence of mental health problems, defined based on a screening instrument (Strengths and Difficulties Questionnaire). Logistic regression models tested factors related to the probability that a child would not present mental health problems at follow-up. RESULTS: The likelihood of a child presenting mental health problems was higher at baseline compared to the two-year follow-up (67.5 and 56.1%, respectively). Absence of mental health problems two years after a psychosocial intervention was significantly correlated with the following baseline factors: lower level of caregiver’s psychiatric symptoms as measured by the SRQ (Self-Report Questionnaire) (AOR = 0.84, p = 0.0065), absence of child physical neglect (AOR = 0.38, p = 0.0705) and parental Protestant religion affiliation, compared to other religions (AOR = 4.06; p = 0.0107). CONCLUSIONS: Different factors are related to the absence of mental health problems of children working on the streets after enrollment in a two-year psychosocial program. Our findings suggest that interventions that aim to improve child mental health should consider the detection of psychiatric symptoms in caregivers, provide treatment when it is needed, and also assess other problems such as neglect in the family setting.