This paper makes use of the substantial information about the psychological and behavioural development of children by age ten in the 1970 Cohort to predict later, economic outcomes, namely qualifications, employment and earnings. It is found that this previously unobserved individual heterogeneity has very substantial implications for the labour market. The returns to education are not significantly reduced by this omission bias but there is evidence of substantial returns to the production of non-academic ability. The paper also finds that different age ten abilities and attributes have implications for different adult outcomes so that human capital production should not be considered by economists as a simple one-dimensional process. Age ten conduct disorder predicts male adult unemployment particularly well but it is self-esteem that predicts male earnings. For women the locus of control variable is particularly important. Finally, whereas age ten maths ability is a good predictor of subsequent educational development for children from high SES families, reading is the stronger predictor for children from low SES groups. The implications of these results for education are developed. Parental attitudes are much more important than raw indices of social class for the explanation of the age ten scores. Schooling curriculum may be important.