International evidence suggests that problem gambling tends to be 2–4 times higher among adolescents as among adults and this proves to be true of Great Britain according to the latest adolescent prevalence survey. 8,958 British children (11–15) were surveyed in 201 schools during late 2008 and 2009. The questionnaire included a standard screen, DSM-IV-MR-J, to test for problem gambling. Our regression models explore influences of demographic, home and school characteristics on probabilities (both unconditional and conditional on being a gambler) of a child testing positive for problem gambling. More than 20% of children participated in gambling and, of these, nearly 8% tested positive. Age-group prevalence of problem gambling was 1.9%, compared with 0.6–0.9% in the most recent official adult surveys. Boys were much more likely than girls to gamble and to exhibit symptoms of problem gambling if they did. Generally, home characteristics, particularly parental attitude and example, dominated school characteristics in accounting for risks. Unanticipated findings included significantly elevated probabilities of problem gambling among Asian children and among children who live in a home without siblings. Child income was also a potent predictor of gambling and problem gambling.