The United Kingdom Childhood Cancer Study (UKCCS) is a national multi-centre case-control study that was designed to evaluate the potential aetiological role of prenatal events in childhood cancer. The obstetric records of 2692 mothers of children diagnosed with cancer and 4864 mothers of children without cancer were available for analysis. Overall, 1754 (65%) case mothers and 3220 (66%) control mothers had at least one prior pregnancy before the birth of the index child. Of these, 12 (0.68%) of the former and 9 (0.28%) of the latter had a prior molar pregnancy (odds ratio 2.5, 95% confidence interval 1.1 - 6.1). Both childhood cancer and molar pregnancy are rare neoplastic events, and the numbers are small. Nonetheless, whilst the associations were strongest for common precursor B-cell acute lymphoblastic leukaemia (OR 5.2, 95% CI 1.9 - 14.7) and sarcoma (OR 6.2, 95% CI 1.3 - 30.3), the spread across the remaining diagnostic groups suggests that the relationship, if confirmed, may be of a generalized, rather than specific, type. This is the first time that an association between childhood cancer and hydatidiform mole has been reported. The UKCCS's systematic use of clinical records permitted a more precise characterization of reproductive events than is possible in investigations that rely on individuals own accounts, and we are confident that our findings cannot be explained by recall bias or other methodological limitations. Accordingly, we suggest that there may be an aetiologic connection between molar pregnancy and childhood cancer, and speculate here on the various genetic/epigenetic mechanisms that could be involved.