This paper uses data from the German Socio-Economic Panel for the years 2000 to 2005 to study the earnings differential between self- and dependent employed German men. Constructing a counterfactual earnings distribution for the self-employed in dependent employment and using quantile regression decompositions we find that the earnings differential over the distribution cannot be explained by differences in endowments. Furthermore, low-earning self-employed could earn more in dependent employment. Finally, the observed earnings advantage for the self-employed at the top of the earnings distribution is not associated with higher returns to observable variables.