Polymorphisms (A33512C, C21151T and PAT -/+) of the xeroderma pigmentosum group C (XPC) were shown to contribute to genetic susceptibility to cancer. However, association studies on these polymorphisms in cancer have shown conflicting results. Thus, we performed a meta-analysis. Overall, there was no significant association between 33512C (9,091 patients and 11,553 controls) and cancer risk. No significant association was found in stratification analysis by tumor sites and ethnicities except an elevated lung cancer risk under the recessive genetic model in all subjects [P = 0.04, odds ratio (OR) = 1.20, 95% confidence interval (CI) 1.00-1.45, P heterogeneity = 0.88]. There was no significant association between 21151T (5,227 patients and 5,959 controls) and cancer risk in all subjects but an increased cancer risk in Caucasians under the recessive genetic model (P = 0.006, OR = 1.45, 95% CI 1.11-1.90, P heterogeneity = 0.75) and homozygote comparison (P = 0.02, OR = 1.41, 95% CI 1.07-1.81, P heterogeneity = 0.41). It might be that 21151T increases bladder cancer risk under the recessive genetic model (P = 0.02, OR = 1.49, 95% CI 1.06-2.09, P heterogeneity = 0.47) and homozygote comparison (P = 0.02, OR = 1.49, 95% CI 1.05-2.11, P heterogeneity = 0.23). There was no significant association between PAT + (4,600 patients and 4,866 controls) and cancer risk in all subjects. An increased cancer risk in Caucasians was found under the recessive genetic model (P = 0.02, OR = 1.20, 95% CI 1.03-1.40, P heterogeneity = 0.37) and homozygote comparison (P = 0.008, OR = 1.26, 95% CI 1.06-1.50, P heterogeneity = 0.13). The XPC PAT + allele might increase head and neck cancer risk (P = 0.02, OR = 1.29, 95% CI 1.04-1.59, P heterogeneity = 0.15). More studies based on larger, stratified, case-control population, especially studies investigate the combined effect of XPC A33512C, C21151T, and PAT, are required to further evaluate the role of these polymorphisms in different cancers.