The purpose of this thesis is to present an autoepistemic approach topreference reasoning. The method controls the generation of extensionsby first linking doxastic representations in a hierarchy,and then by ensuring the maintenance of integrity between the levelsin the hierarchy using so-called integrity constraints. This approachis based on an encoding of ordered default theories into an ''onlyknowing'' logic with confidence levels, and is the result of studyingseveral principles for default logic approaches topreference, as well as a critique of the hierarchic autoepistemiclogic of Konolige with respect to these same principles. We show thatour approach is the first autoepistemic formalism that handlespreference according to all the principles mentioned above. Anothermain result of our work is that the generation of preferred extensionsmay be carried out entirely at the object level. Thus, as opposed toapproaches to preference using default logic, our approach provides aconstructive method for determining extensions of ordered default theories.