Background: Street children are at risk of various high risk behaviors, but few studies have been performed in this field in Iran and worldwide. Objectives: This study aimed to assess the prevalence of high risk behaviors and its predictors among street children to design interventions and future research agenda regarding these children in Iran, especially in southeast region. Patients and Methods: This cross-sectional study was conducted in southeast Iran in 2015. Based on postal map, the city was divided into 5 districts and 216 male street children were selected using snowball sampling. An unstructured questionnaire was provided by integrating 8 previously used questionnaires in different studies with approved validity and reliability and filled through interview and observation. Data was analyzed in Stata.12 software using bivariate and multiple logistic regression by Hosmer-Lemeshow method. Results: The mean age of subjects was 12.5 ± 3.2 years. The lowest mean age of initial use and usually firstly used substance was for cigarette smoking. The prevalence of high risk behaviors was 47.7% (95% CI: 44% - 51%). In final multiple logistic regression model, separation of parents (OR: 3.8 [95% CI: 1.86 - 16.2]), peddling activity type (OR: 2.73 [95% CI: 1.28 - 9.5]) and lower fathers’ educational level (OR: 3.83 [95% CI: 1.86 - 15.4]) significantly increased the prevalence of high-risk behaviors among street children. Conclusions: The prevalence of high-risk behaviors among street children in southeast Iran is almost high. In this regard, street vendors and children who their parents are illiterate or have been separated are more at risk. Generally, principal organizing these children and family education could decline their problems. However, these actions need intersectoral collaboration in society.