The German educational system is characterized by a large sector of dual vocational training, which facilitates integration into the labour market. This system creates a specific training market for school leavers, which is characterized by strong regional disparities. These differences as well as their consequences have not been systematically analysed in previous research. In a theory-guided analysis this paper examines empirically which structural 'handicaps' affect regional transition rates from school to training and how regional training markets may be classified according to these structural factors. To this end, a new method is applied which combines regression and cluster analysis to avoid arbitrariness in the selection of classification variables. It generates a well-interpretable classification of vocational education markets, which is of broad use in research and labour market policy. The method may be applied to solve a broad variety of similar research problems in regional science. JEL Classification I21 · J24 · R23Electronic supplementary material The online version of this article (https://doi.org/10.1007/s00168-017-0856-z) contains supplementary material, which is available to authorized users.