The investigation of an association between increased exposure to residential extremely-low frequency electromagnetic fields (ELF-EMF) and childhood leukemia was part of a population-based case-control study carried out between 1992 and 1995 in the northwestern part of Germany. A total of 129 children with leukemia and 328 controls participated in the EMF-study. Exposure assessment comprised measurements of the magnetic field over 24 hours in the child's bedroom at the residence where the child had been living for the longest period before the date of diagnosis, and spot measurements at all residences where the child had been living for more than one year. The median of the 24h-measurement in the child's bedroom was regarded as the most valid exposure variable. For children exposed to more than 0.2 microT, an elevated but not significant odds ratio (OR) was observed (OR = 3.2, 95 percent confidence interval = 0.7-14.9). These figures are based on only four leukemia cases and three controls since only 1.5 percent of the study population was classified as highly exposed. Exploratory analyses revealed ORs that were not statistically significantly increased for other characteristics of the magnetic field at varying cut-points. The results are comparable with those from other studies. Although not statistically significant, they may indicate a positive association between EMF and childhood leukemia.