This paper tests for the impact of school size, measured by number of pupils, on academic performance at age 16 controlling for exam scores 2 years earlier. Schools of more than at the most 600 pupils tend to achieve poorer public exam results in the long run and a higher proportion fail to gain any qualifications. School size affects attendance rates as well, and since attendance contributes to exam outcomes, there is an additional small indirect impact of size. Larger classes are also associated with poorer results. Schools offering post-compulsory education for 16-18 year olds perform less well than those restricted to the 11-16 age range, consistent with excessive resources allocated towards the older students. In response to better exam results, schools slightly increase in size but there is no tendency for schools with higher proportions of deprived pupils to contract.