Whether the powerful medications used to treat epilepsy increase the risk of cancer has been debated for decades, but until now no study could disentangle the contributions of anti-epileptic medications and epilepsy itself to cancer risk. Using a cohort comprising all Danish residents ≥16 years old at some point during the period 1996-2010 (>56 million person-years of follow-up) and information from national health registers, we examined associations between anti-epileptic medication use and cancer rates in persons with and without epilepsy, and between epilepsy and cancer rates in treated and untreated individuals. Associations were expressed as incidence rate ratios (IRRs) estimated using Poisson regression. Among persons without epilepsy, use of anti-epileptic medication increased the rates of most cancers little or not at all, although we observed moderately increased rates of liver, mouth and throat, and respiratory tract cancers (IRRs 1.40-1.59). In contrast, we observed strong associations between epilepsy and the rates of central nervous system and mouth and throat cancers (IRRs 2.00-3.91), and a modest association between epilepsy and the rate of respiratory tract cancers (IRRs 1.30-1.35), independent of anti-epileptic medication use. Our finding of only modest increases in cancer risk directly attributable to anti-epileptic medication use suggests that these medications may not be as strongly carcinogenic as has been feared, and that it is not primarily anti-epileptic medications that are responsible for the increased cancer risk among epileptics but another aspect of epilepsy diagnosis or treatment or an etiologic factor common to the two conditions. What's new? Do anti-epileptic medications increase the risk of cancer? The question has been debated for decades with no clear answer. In this study, the authors used a large Danish registry to tease out the contribution of medication use versus epilepsy alone to cancer rates. Their findings indicate that anti-epileptic medications may not be as strongly carcinogenic as has been feared, and suggest that some aspect of epilepsy itself - or perhaps an etiologic factor common to both epilepsy and cancer - is responsible for the previously observed association between these medications and increased cancer risk.