Most reviews on the carcinogenicity of formaldehyde have focused on cancers of the respiratory tract because these cancer sites were thought to be most biologically plausible. However, two recent updated large industrial studies have found positive associations between some measures of formaldehyde exposure and increased leukemia, although another recent update of a large study did not find any association. We examine 18 epidemiology studies of workers exposed to formaldehyde where leukemia rates were reported. We summarize the findings across studies by using meta-analytic techniques to calculate the meta-relative risk values (mRR), confidence intervals, and heterogeneity of the risk estimates for several study characteristics. We also determine if publication or reporting biases may be affecting the estimates. We found a small increase in rate of leukemia overall among embalmers (mRR = 1.6, 95%CI 1.2-6.0), and pathologists/anatomists (mRR = 1.4, 95%CI 1.0-1.9). Industrial workers, who have been reported to have the highest formaldehyde exposures, had a mRR of 0.9 (95%CI 0.8-1.0). There was increased risk with increasing exposure in two large industrial studies, although the increased risk is one of these studies was not seen when an external comparison group is used. Also, another large industrial study with more highly exposure workers found decreased risk of leukemia among the highest exposed group. The long latency for leukemia deaths observed in the two industrial studies that reported increased risk was not consistent with a chemical carcinogen such as benzene. We found limited evidence of publication or reporting bias. On balance, these data do not provide consistent support for a relationship between formaldehyde exposure and leukemia risk.