We review the epidemiological evidence on childhood leukemia and residential exposure to 50/60 Hz magnetic fields. The possibility of carcinogenic effects of power frequency magnetic fields (ELF-EMF), at levels below units of micro tesla (microT), was first raised in 1979 by a case-control study on childhood cancer carried out in Denver, USA. In that study, excess risks of total cancer and leukemia were observed among children living in homes with "high or very high current configuration", as categorised on the basis of proximity to electric lines and transformers. Many other epidemiological studies have been published since then, characterised by improved--although still not optimal--methods of exposure assessment. At the end of 2000, the epidemiological evidence to support the association between exposure to extremely-low-frequency magnetic fields and the risk of childhood leukemia is less consistent than what was observed in the mid 90s. At the same time, a growing body of experimental evidence has accumulated against both a direct and a promoting carcinogenic effect of ELF-EMF. Such "negative" experimental evidence hampers a causal interpretation of the "positive" epidemiological studies.