Policy learning is an important factor of policy change and designates the relatively enduring alterations of policymakers’ beliefs, attitudes and behavioral intentions with respect to policies. While there is a considerable amount of knowledge about the influence of institutions, organizations, and networks on policy learning, little research has been conducted about the individual conditions of this mechanism. Based on an analytical framework combining the Advocacy Coalition Framework’s concepts (at the collective level) and the Theory of Reasoned Action’s concepts (at the individual level), this research focuses on the individual conditions of policy learning, in particular policymakers’ interests and trust. The test of the research hypotheses is mainly based on the analysis of the data from a web survey conducted among 1256 policymakers from the Belgian rail and electricity sectors. The use of multilevel regression models to analyse the data of the survey allows to control the influence of aggregate-level effects and to focus on the individual-level conditions of policy learning. The research focuses on the evolution of policymakers’ preferences with respect to a policy (here, the European liberalization policy process) as an operational measure of policy learning. The research shows that policymakers’ preferences do not evolve very much over time. They do evolve, however. To explain this, key factors are policymakers’ perceptions about the impacts of the policy on their personal interests and the interests of their organization: policymakers will adapt their policy preferences according to these perceptions. Nevertheless, much room remains for other determinants of policy learning. This includes socio-psychological factors, like the climate of trust that policymakers perceive in the policy subsystem in which they are involved. This climate can prompt them to recognize more easily the negative effects that policy changes have on them and consequently inclines their policy preferences to evolve ...