Purpose of this study was investigating the effect of emotional intelligence training on aggression of aggressive adolescences. Research method was qui-experimental (pre test-post test method) with control group. For this purpose, at first from five district education guidance schools on Mashhad city, guidance school selected by clustered random sampling method and 100 students selected randomly. Then they answer to Nelson anger questionnaire (2000), so 20 students that get top scores, screened as aggressive and categorized in two "experimental" and "control" group randomly. Students of both groups has completed the "Schutte emotional intelligence questionnaire" and also "Nelson's anger inventory", and then emotional intelligence trained to "experimental" group in 10 session. Data's analyzed by "SPSS" software, t-test, Pierson correlation coefficient and simple Linear Regression. Results of pierson correlation coefficient show that there is a significant negative correlation between emotional intelligence and aggression in aggressive adolescences (r = -0.562). After emotional intelligence training, there is a meaningful difference between two groups from emotional intelligence view point and its parameter, included: regulating, assessment and statement and using emotional (p