BACKGROUND: We investigated whether low birth weight constitutes a causal risk factor for child problem behavior, using a variation of the co-twin control method. METHODS: In a representative sample of 745 twin pairs (monozygotic: 324 pairs), birth weight was recorded at birth and child problem behavior at mean age 10 years was measured with the Child Behaviour Checklist (CBCL). RESULTS: Lower birth weight was a continuous risk factor for later child problem behavior (adjusted regression coefficient over units of 500 g: beta = -.15, p =.046), and greater levels of within-pair CBCL discordance did not result in a reduced effect size. Greater within-pair birth weight discordance was associated with greater within-pair CBCL score discordance (beta =.35, p <.001). This latter effect was similar in monozygotic (beta =.34, p =.005) and dizygotic twins (beta =.37, p =.003). CONCLUSIONS: The fact that (1) the effect size of the association between low birth weight and child problem behavior was not reduced in pairs with greater levels of CBCL discordance, and (2) similar effect sizes were found in monozygotic and dizygotic twins for the within-pair association between birth weight discordance and CBCL score discordance, suggests that the observed relationship between low birth weight and child problem behavior is not due to a shared environmental or genetic variable that influences both characteristics. Lower birth weight is a causal risk factor for child problem behavior, the effects of which may well extend into adulthood.