The role of natural background radiation (NBR) in childhood acute leukemia (AL) remains unclear. Several large record based studies have recently reported heterogeneous results. Differences in exposure assessment timing may explain this heterogeneity. In a previous ecological study we did not observe any association between childhood AL incidence in France and NBR exposure at the time of diagnosis. With the same methodology, the present study focused on NBR exposure at the time of birth. Based on data from the French national registry of childhood cancer, we analyzed all AL together, and lymphoblastic and myeloid AL, separately. We included 6,059 childhood AL cases born and diagnosed in mainland France between 1990 and 2009. NBR levels in municipalities of residence at birth were estimated by cokriging models, using NBR measurements and precise geological data. The incidence rate ratio (IRR) per unit variation of exposure was estimated with Poisson regression models, with adjustment for socio-demographic indicators and ultraviolet radiation levels. NBR exposures were considered at the time of birth, and cumulatively from birth to diagnosis. We also estimated a total NBR dose to red-bone marrow (RBM). There was no evidence for an association between NBR exposure at birth and childhood AL incidence, neither overall (gamma radiation: IRR = 0.99 (0.94,1.05) per 50 nSv/h; radon: IRR = 0.97 (0.91,1.03) per 100 Bq/m3) nor for the main AL types. The conclusions were similar with the cumulative exposures, and the total RBM dose. The study was based on high quality incidence data, large numbers of AL cases, and validated models of NBR exposure assessment. In all, the results further support the hypothesis that NBR are not associated to childhood AL in France. •The initiation process of childhood leukemia might begin during the perinatal term.•Nationwide study on background radiation exposure at birth and childhood leukemia.•Complete registration of childhood leukemia in France (6,059 cases over 1990–2009).•Cockriging models based on radiation measurements (>10,000 sites) and geological maps.•Exposure to background radiation at birth not associated with childhood leukemia.