This study aimed to identify immune-based prognostic biomarkers associated with metastasis of osteosarcoma. Based on the GEO and TCGA databases, 437 differentially expressed genes were screened between primary and metastatic osteosarcoma. Weighted gene co-expression network analysis (WGCNA) revealed 496 genes in turquoise module which had the highest correlation with osteosarcoma metastasis. Within these two group genes, 122 common genes involved in osteosarcoma metastasis were identified. These genes were enriched in chemokine activity, chemokine receptor binding, TNF signaling pathway, etc. Survival analysis revealed 8 prognostic genes (ANK3, EGR1, FBP1, FOS, KIFC3, MAOB, ISLR and MFAP4) from the 122 genes. RT-qPCR showed that all of these eight genes were differentially expressed between 143B and MNNG/HOS Cl cells. Various infiltrating immune cells showed significant differences between primary and metastatic osteosarcoma. Expression of all the 8 prognostic genes was correlated with infiltration abundance of multiple immune cells, such as follicular helper T cells, activated dendritic cells. In addition, 10 microRNAs and 7 transcription factors that targeted these prognostic genes were predicted. In conclusion, 8 immune-based prognostic genes associated with osteosarcoma metastasis were identified.