Logistics costs after natural disasters may be reduced through the implementation of reliability models in the design stage of distribution networks. This research compares two mathematical formulations for the reliable uncapacitated facility location problem (RUFL) which model correlated disruptions in order to minimize the sum of fixed and transportation costs. The study compares a RUFL model which reduces unsatisfied service by allowing the decision maker to price unsatisfied demand using a penalty cost and a RUFL model with service levels (RUFL-SL) that ensures a minimum level of satisfied demand for each scenario. Experimental results with correlated random disruptions suggest that the RUFL-SL dominates relevant solutions of the RUFL under some experimental conditions, but is dominated by the RUFL under other experimental conditions. It is concluded that the RUFL-SL model is preferable to the RUFL when service interruptions may not be easily translated into penalty costs or when service is critical.