We present an algorithm that is able to integrate uncertain probability statements of different default levels. In case of conflict between statements of different levels the statements of the lower levels are ignored. The approach is applicable to inference networks of arbitrary structure including loops and cycles. The simulated annealing algorithm may be used to derive a distribution which best fits to the different statements according to the maximum likelihood principle. In contrast to Pearl's approach to probabilistic default reasoning based on probabilities arbitrarily close to 1 our approach may combine conflicting evidence yielding a compromise between statements of the same default level according to their relative reliability. Between observationally equivalent solutions the maximum entropy criterion is employed to select a distribution with minimal higher order interactions.