The X-ray repair cross-complementing group 3 (XRCC3) is a highly suspected candidate gene for cancer susceptibility. However, association studies on the XRCC3 polymorphisms (4541A>G, Thr(241)Met, 17893A>G) in cancer have shown conflicting results. Therefore, we performed a meta-analysis to better assess the purported associations. Forty eight eligible case-control studies including 24,975 cancer patients and 34, 209 controls were selected for our meta-analysis. Overall, individuals carrying the XRCC3 Met/Met genotype showed a small cancer risk under a recessive genetic model. The subgroup and meta-regression analysis demonstrated different scenarios concerning the XRCC3 Met/Met genotype's role in cancer susceptibility for different subgroups. Specially, there was a significantly increased risk of breast cancer (OR, 1.14; P=0.0004; 95% CI, 1.06-1.23; P=0.37 for heterogeneity), elevated but not significant risk of cancer for head and neck, bladder, surprisingly, a significantly decreased risk of non-melanoma skin cancer (OR, 0.76; P=0.007; 95% CI, 0.62-0.93; P=0.61 for heterogeneity). A significantly elevated risk of cancer was observed in population-based case-control studies but not in nested or hospital based studies. Similarly, we found a significantly increased risk of cancer for A4541G and a decreased risk for A17893G under dominant genetic models. Our meta-analysis results support that the XRCC3 might represent a low-penetrance susceptible gene especially for cancer of breast, bladder, head and neck, and non-melanoma skin cancer. A single larger study should be required to further evaluate gene-gene and gene-environment interactions on XRCC3 polymorphisms and tissue-specific cancer risk in an ethnicity specific population.