This study uses data from the British National Child Development Survey (NCDS) to examine interactions between socio-economic status (SES), children's test scores, and future wages and employment. We find that children of lower SES have both lower age 16 test scores and higher returns to these test scores in terms of age 33 wages and employment probabilities than high-SES children. We then examine determinants of age 16 scores. Conditional on having had the same age 7 mathematics scores, high-SES children go on to achieve higher age 16 mathematics scores than children of low or middle-SES. They are also much more likely to pass O-levels in English and Mathematics. These differences are either eliminated or greatly reduced when observable measures of school quality are added to the model, suggesting that high-SES children get better age 16 test scores at least in part because they attended better schools. On the other hand, conditional on age 7 scores, low-SES children achieve higher age 16 reading scores than high-SES children and the estimated relationship between the two is not affected by the addition of school quality variables. This observation provides evidence consistent with the conjecture that success in reading may be less dependent on school quality than success in mathematics.