After identifying an association between maternal Epstein-Barr virus (EBV) reactivation and acute lymphoblastic leukemia (ALL), the authors analyzed a nested case-control study within Finnish and Icelandic maternity cohorts with 7 million years of follow-up to confirm EBV's role in ALL. Offspring of 550,000 mothers were followed up to age 15 years during 1975-1997 by national cancer registries to identify leukemia cases. Mothers of cases and three quarters of matched mothers of controls were identified by national population registers. First-trimester sera from mothers of 304 ALL cases and 39 non-ALL cases and from 943 mothers of controls were analyzed for antibodies to viral capsid antigen, early antigen, and EBV transactivator protein ZEBRA. Relative risk, estimated as odds ratio (95% confidence interval), was adjusted for birth order and sibship size. Combining early antigen and/or ZEBRA immunoglobulin G antibodies with the presence of viral capsid antigen immunoglobulin M antibodies did not increase the estimate for ALL risk for viral capsid antigen immunoglobulin M alone (odds ratio = 1.9, 95% confidence interval: 1.2, 3.0). Both ZEBRA immunoglobulin G antibodies and viral capsid antigen immunoglobulin M antibodies were associated with an increased risk of non-ALL in the offspring (odds ratio = 4.5, 95% confidence interval: 1.3, 16; odds ratio = 5.6, 95% confidence interval: 1.1, 29, respectively), suggesting EBV reactivation in the mothers of non-ALL cases. EBV reactivation may be associated with a proportion of childhood leukemia.