Traditional reliability enhancement programs in power distribution utilities are focused on averaging the reliability improvement for distribution utility customers. The provision of targeted reliability levels at optimized operation costs for individual distribution utility customers is an important feature of smart distribution systems that have been of interest recently. Consequently, a market-based approach is required for identifying the minimum cost of distribution system operation that considers customers’ reliability preferences (outage costs) in a competitive system operation. In this paper, we introduce a proposed market-based approach for setting customer reliability preferences in a smart distribution system. The proposed approach utilizes the game theory as a powerful tool for modeling conflicts among distribution system participants (customers and the utility). In this model, customers will propose their desired reliability level and the distribution system operator makes reliability allocations among customers which minimize the distribution system operation cost. The interactions among participants are modeled and solved using a bi-level optimization approach to determine the optimal customer reliability level. The model can be utilized for operating the existing network or in the case where a utility would invest on the reliability enhancement. The proposed model can suitably encourage the participation of distribution system customers in reliability enhancement by offering several reliability options via two-way communications between customers and the distribution system utility. The simulation tests on a sample distribution system demonstrate the significant features of the proposed distribution reliability model.