Our objective was to test the hypothesis that the risk of childhood leukemia is associated with allergies or a family history of allergy. We used a German population-based case-control study with self-reported information on allergies of the children and their first-degree relatives. Our study included a total of 1,130 cases of acute lymphoblastic leukemia (ALL), 164 cases of acute myeloid leukemia (AML) and 2,957 controls. A major finding of our study is that hay fever, neurodermatitis and contact eczema are underrepresented within the group of children with ALL, with respective odds ratios (OR) of 0.45 (95% confidence interval [CI] 0.31–0.66) for hay fever, of 0.49 (CI 0.34–0.71) for neurodermatitis and of 0.62 (CI 0.39–0.99) for eczema, respectively. Atopic diseases, comprising hay fever, neurodermatitis and asthma, are much stronger related with a reduced risk of ALL than other allergies (OR 0.52, CI 0.40–0.67 vs. OR 0.89, CI 0.66–1.21). The strongest association is seen with an atopy in the index child; however, ALL risk is also reduced if one of the parents or a sibling had an atopic disease. No such consistent pattern is seen for AML. Our data suggest that atopy or a family history of atopy are associated with a reduced risk of childhood ALL. Recall bias remains a concern, but sensitivity analysis provided some evidence that the protective effect is unlikely to be attributable to this bias in its entirety. © 2003 Wiley-Liss, Inc.