This analysis tests the assumptions of the principal-agent model on delegation in EU conciliation processes and discusses the implications of the empirical results for the value of the model as a general framework of delegation. It shows that irresponsible behaviour by the EU conciliation committee is rare, contrary to the usual prediction of the principal-agent framework, because a series of the common assumptions of the model are not fulfilled. It is argued that, in a number of cases, the principal-agent model may still be relevant if some of its assumptions are endogenised into the model as characteristics of the environment, but that there may be other types of delegation where the core assumptions of the model do not apply. The findings are generated on the basis of different data sources such as official documents from the EU institutions, descriptive statistics and semi-structured interviews.