Although exposure to high levels of arsenic in drinking water is associated with excess cancer risk (e.g., skin, bladder, and lung), lower exposures (e.g., <100-200 microg/L) generally are not. Lack of significant associations at lower exposures may be attributed to methodologic issues (e.g., inadequate statistical power, exposure misclassification), or to differences in the dose-response relationship at high versus low exposures. The objectives of this review and meta-analysis were to evaluate associations, examine heterogeneity across studies, address study design and sample size issues, and improve the precision of estimates. Eight studies of bladder cancer and low-level arsenic exposure met our inclusion criteria. Meta-analyses of never smokers produced summary relative risk estimates (SRREs) below 1.0 (highest versus lowest exposure). The SRRE for never and ever smokers combined was elevated slightly, but not significantly (1.11; 95% CI: 0.95-1.30). The SRRE was somewhat elevated among ever smokers (1.24; 95% CI: 0.99-1.56), and statistical significance was observed in some subgroup analyses; however, heterogeneity across studies was commonly present. Although uncertainties remain, low-level arsenic exposure alone did not appear to be a significant independent risk factor for bladder cancer. More studies with detailed smoking history will help resolve whether smoking is an effect modifier.