Objective: Currently published data regarding the potential role of osteoprotegerin (OPG), osteocalcin (OCN) and osteopontin (OPN) for the discrimination between rheumatoid arthritis (RA) and osteoarthritis (OA) are contradictory. To derive a more precise evaluation, a meta-analysis was performed. Methods: Published literatures comparing plasma/serum OPG, OCN and OPN levels between RA group and OA controls were searched in PubMed, Embase and the Cochrane Library. The Newcastle-Ottawa Scale was used to assess the study quality. Pooled standard mean difference (SMD) with 95% confidence interval (CI) was calculated by random-effect model analysis. Heterogeneity test was performed by the Q statistic and quantified using I². Results: Nine studies including 438 RA patients and 255 OA patients were finally incorporated in the meta-analysis after examining title, type, abstracts and full text. The results showed that RA patients had higher plasma/serum OPN (pooled SMD = −2.57, 95% CI = −4.72 to −0.41) levels when compared to OA patients. No significant difference in plasma/serum OPG (pooled SMD = −0.29, 95% CI = −1.07‒0.49) and OCN (pooled SMD = −0.09, 95% CI = −0.48‒0.31) levels were found between RA patients and OA patients. Subgroup analysis indicated that plasma/serum OPG levels had no significant differences between RA patients and OA patients in Europe and Asian. Conclusions: Overall, there is no significant difference in circulating OPG and OCN levels between RA patients and OA patients. However, plasma/serum OPN level is significantly higher in RA patients compared with OA patients.