The objective of this research was to compare the association of birthweight alone with gender-specific birthweight-for-gestational age on childhood cancer risk in a large population-based case-control study in Germany. Incident cases of childhood cancer (n=2,024, diagnosed 1992-1994) were ascertained from the German Childhood Cancer Registry. Controls were randomly drawn from population registries. Parents reported risk factor information in a mailed questionnaire and telephone interview. The odds ratio for acute lymphoblastic leukemia (ALL) was 1.41 (95% confidence interval 1.08-1.84) in the high-birthweight category (>4 kg) and was 1.45 (1.07-1.97) in the large-for-gestational age (LGA) category compared to the normal birthweight (2.5-4 kg) and the appropriate-for-gestational age (AGA) categories, respectively. However, the agreement between the birthweight and birthweight-for-gestational age was only moderate. Subgroup analyses revealed elevated odds ratios for ALL and CNS tumors in first born's who were LGA but of normal birth weight. Thus, two findings from this post-hoc analysis are worthy of note: (1) the use of birthweight-for-gestation age categories within birthweight sub-groups potentially identified new high-risk groups among firstborns for ALL tumors and among all children for CNS tumors; and (2) although the magnitudes of risk estimators for ALL were comparable in the traditional high-birthweight group and in the LGA, the same children were not jointly classified in the same newborn categories indicating two potentially different subsets of children at risk.