Historically, reliability analysis has ignored the occurrence of natural hazards such as those associated with extreme weather, flooding, earthquakes and tsunamis, which are becoming more and more relevant due to recent events. In this context, we present an assessment methodology to determine the resiliency levels of a distribution system exposed to a major earthquake. The proposed 4-stage methodology models (i) the earthquake, (ii) the fragility of network components, (iv) the network outages, and (v) the impacts on network operation (including the energy not supplied) through sequential Monte Carlo simulations. This methodology is used to study the resilience of distribution networks under two particular strategies: one that hardens substations infrastructure in order to reduce their fragility levels, and the other one that uses additional network infrastructure in the form of transfer cables to shift load among substations in case a major event occurs. Through several case studies based on a real distribution network in Chile, we demonstrate that hardening substations infrastructure may be a more resilient way to deal with earthquakes, even when compared to an alternative, extreme case where a vast number of transfer cables are installed to support damaged substations.