This paper estimates the effects of an education policy (Triple Science) in England aimed at increasing the take-up and attainment of young people in science subjects. The effect of the policy is identified by comparing two adjacent cohorts of pupils in schools that offer Triple Science to one cohort, but not to the other. The results suggest some large and significant effects on later subject choice and attainment, and these appear to be particularly strong for boys and pupils from more deprived backgrounds.