The number of children born via Assisted Reproductive Technology (ART) has significantly increased over time, yet we know relatively little about the longer term outcomes for these children. The bulk of evidence shows that ART is linked to higher risks of adverse health outcomes at birth, and there is mixed evidence as to whether ART children perform better over time, with some studies showing higher cognitive development compared to naturally conceived children. We test whether these mixed findings are related to selective characteristics of ART parents. ART mothers are a selective group and on average older, better educated, and more likely to be employed and married. Using data from the first 5 waves of the UK Millennium Cohort Study (N=13,647), this paper compares the verbal cognitive development (as measured by the British Ability Scale) of children born via ART with naturally conceived children up to the age of 11 in 2000-1. Findings from growth curve models show that ART children perform higher than average, but that this effect disappears when controlling for parental background. Further, we show that differences in cognitive abilitie