OBJECTIVES: This study uses meta-analysis methodology to examine the statistical consistency and importance of random variation among results of epidemiologic studies of occupational electromagnetic field exposure and leukemia. METHODS: Studies for this meta-analysis were identified from previous reviews and by asking researcher active in this field for recommendations. Overall, 27 studies of occupational electromagnetic field exposures and leukemia were reviewed. A variety of meta-analysis statistical methods have been used to assess combined effects, to identify heterogeneity, and to provide a single summary risk estimate based on a set of simiar epidemiologic studies. In this study, classification of exposure metircs on occupational epidemiologic studies are reported for (1) job classification (20 individual studies); (2) leukemia subtypes (13 individual studies); and (3) country (27 individual studies). RESULTS: Results of this study, an inverse-variance weighted pooling of all the data leads to a small but significant elevation in risk of 11% (OR=1.11, 95% CI : 1.06~1.16) among 27 occupational epidemiologic studies. Publication bias was assessed by the 'fail-safe n' that may be not influence for all combined results exception a few categories, ie, "power station operators" and "electric utility workers" by job classification on occupational study. And all combined odds ratio results were similar for fixed-effects models and random-effects models, with slightly higher risk estimates for the random-effects model in situations where there was significant heterogeneity, ie, Q-statistic significant (p<.05). CONCLUSIONS: We found a small elevation in risk of leukemia, but the ubiquitous nature of exposure to electromagnetic fields from workplace makes even a weak association a public health issue of substantial power to influence the present overall conclusion about relationship between electromagnetic fields exposure and leukemia.