We introduce a translation of a conditional logic semantics to a mathematical programming problem. A model of 0-1 programming is used to compute the logical consequences of a conditional knowledge base, according to a chosen default theory semantics. The key to understanding this model of mathematical programming is to regard the task of the entailment of plausible conclusions as isomorphic to an instance of weighted MAX-SAT problem. Hence, we describe the use of combinatorial optimization algorithms in the task of defeasible reasoning over conditional knowledge bases.