Osteosarcoma (OS) is the most common bone cancer, mainly diagnosed in children and adolescents. So far, no reliable molecular biomarkers have been identified to effectively evaluate OS prognosis and immune infiltration. Herein, we curated transcriptome profiles and clinical information from the publicly available OS cohorts to establish an immune-related prognostic signature. Besides, immunotherapeutic cohorts of urothelial cancer and melanoma patients were also employed to infer immunotherapy prediction roles of the identified signature. Lymphocytes infiltration, immune response-related pathways and signatures in the microenvironment were assessed according to distinct risk subgroups. Based on the univariate Cox analysis and further feature selection implemented by the LASSO regression model in the TARGET cohort, a 21-immune-gene signature was identified by combing the expression values and corresponding coefficients. We observed that the low-risk score of this signature was significantly linked with the preferable survival outcome (Log-rank test P < 0.001). The consistent results of better prognoses of the low-risk group were also obtained in subsequent two validation cohorts. Immunology analyses showed that favorable immune infiltration and elevated enrichment of immune response signals may contribute to the better outcome of the low-risk OS subgroup. The immunotherapeutic efficacy analyses demonstrated that low-risk patients harbored significantly enhanced response rates and improved immunotherapy survival outcomes. Together, our established signature could evaluate survival risk and represent the immune microenvironment status of OS, which promotes precision treatment and provides a potential biomarker for prognosis prediction and immunotherapy efficacy assessment.