The role of education in selection and allocation processes on the labour market has been a major research topic for several decades. Surprisingly, the concept of education has been operationalised quite poorly. Usually only a measure of the level of education is used, which is of course not sufficient to understand the full importance of education. Furthermore, conventional regression analysis tends to underestimate the impact of education, since it takes only into account the measured effects of education (such as the effect of level of education) and not the unmeasured effects. In this article we apply multi-level analysis to estimate the systematic effect of education. Dutch school-leaver surveys are used, which consist of more than 20,000 respondents covering more than 200 types of education. We focus on three labour market aspects: the odds of obtaining a permanent job, the odds of obtaining a marching job and the wages. The results show that part of the variation in labour market success between types of education is simply the effect of the individual characteristics gender, social background, ethnicity and age. In addition however, three different components of education itself (specificity, selectivity and complexity) affect the labour market position. Finally, supply/demand differences in the specific segment of the labour market for which the particular type of education prepares and regional variation play an important role. [ABSTRACT FROM AUTHOR]