Previous studies showed that human leukocyte antigen (HLA)-G is specifically upregulated in renal cell carcinoma (RCC). However, a larger cohort of RCC patients are necessary to obtain more information. In this study, 109 RCC primary lesions (clear cell, n = 95; chromophobe, n = 4; papillary, n = 4; collecting duct, n = 6) and corresponding adjacent tumor-negative renal tissues (n = 34) were analyzed for the HLA-G expression by immunohistochemistry (IHC). Meanwhile, plasma soluble HLA-G (sHLA-G) from 16 RCC patients and 144 sex- and age-matched normal individuals was detected by enzyme-linked immunosorbent assay. Correlations between lesion HLA-G expression and various clinical parameters were evaluated. Receiver-operating characteristic (ROC) curve analysis was used to determine the feasibility of HLA-G protein staining and sHLA-G as a diagnosis marker for RCC. IHC data showed that HLA-G was observed in 49.5% of clear cell, 50% of either chromophobe or collecting duct RCC lesions but undetectable in papillary RCC and tumor-negative renal tissues. This finding was consistent with the western blot results. sHLA-G was pronouncedly increased in RCC patients when compared with normal controls (median: 39.5 vs 19.2 U/ml, P = 0.002). However, no correlation was observed between HLA-G expression and various clinical parameters. We found that the area under ROC curve for HLA-G expression and sHLA-G was 0.739 [95% confidence interval (95% CI): 0.659-0.816, P = 0.000] and 0.733 (95% CI: 0.619-0.847, P = 0.002), respectively. Our findings indicated that, except the papillary RCC, other types of RCC could express HLA-G. Furthermore, both lesion HLA-G expression and plasma sHLA-G level might be a useful preoperative biomarker for diagnosis.