To investigate correlations between excision repair cross-complementation group 1 (ERCC1) and 2 (ERCC2) polymorphisms and osteosarcoma prognosis, we conducted a meta-analysis of studies published through October 2016. Studies were identified in the PubMed, ScienceDirect, Springer, and Web of Science databases using preferred reporting items for systematic reviews and meta-analyses (PRISMA). Odds ratios (ORs) or hazard ratios (HRs) and their 95% confidence intervals (CIs) for overall survival (OS), tumor response (TR), and event-free survival (EFS) were estimated. Our meta-analysis included eleven studies in which four SNPs (ERCC1 rs11615 and rs3212986, ERCC2 rs13181 and rs1799793) reportedly associated with osteosarcoma prognosis were investigated. Each of these studies scored > 6 on the Newcastle-Ottawa Scale (NOS). We found that only one SNP, ERCC1 rs11615, correlated with improved OS and TR. The HR of T vs. C for OS was 1.455 (T/C, 95% CI = 1.151–1.839, P = 0.002, I2 = 37.80%). The OR of T vs. C for good TR was 0.554 (T/C, 95% CI = 0.437–0.702, P < 0.001, I2 = 0%). Few significant outcome was observed in subgroup analyses stratified based on study characteristics with adjustments for potential confounders. Our results suggest that ERCC1 rs11615 CC is associated with a better clinical outcome. This suggests rs11615 may be a useful genetic marker for predicting osteosarcoma prognosis.