This study investigates selective participation by area deprivation in two population-based, nationally-representative surveys of children aged 5–16 years in England. We analysed the English participants (N = 22,903) in the British Child and Adolescent Mental Health Surveys, conducted in 1999 and 2004. The surveys recruited parents, teachers and children, allowing us to calculate participation rates for all three types of informant, with particularly detailed analyses possible for parents. Area-deprivation was assigned using Indices of Multiple Deprivation, 2004. We found that parents, teachers and children from more deprived areas were all substantially less likely to participate, with participation 8–9% lower in the most deprived areas than in the least deprived areas. There was substantial heterogeneity in the effect of deprivation on different types of parental non-participation, with a strong association for overall non-contact but none for overall non-cooperation. Refusal to participate in response to opt-out letters showed the opposite gradient to all other forms of non-participation, with individuals from less deprived areas being more likely to opt out. These findings indicate that the deprivation level of area of residence predicts non-response in multiple types of informants, which may exacerbate bias when estimating the prevalence of health outcomes. Furthermore, the magnitude and the direction of this may depend on the recruitment method used. Differential response by area deprivation may therefore be worth measuring in any survey.