Firms that rely on functioning mission-critical equipment for their businesses cannot afford significant operational downtime due to system disruptions. To minimize the impact of disruptions, a proper incentive mechanism has to be in place so that the suppliers provide prompt restoration and recovery services to the customer. A widely adopted incentive mechanism is performance-based contracting (PBC), in which suppliers receive compensation based on realized system uptime. A key obstacle is that disruptions occur infrequently, making it very expensive for a supplier to commit the necessary resources for recovery because they will be idle most of the time. In this paper, we show that designing a successful PBC creates nontrivial challenges that are unique to this environment. Namely, because of the infrequent and random nature of disruptions, a seemingly innocuous choice of performance measures used in contracts may create unexpected incentives, resulting in counterintuitive optimal behavior. We compare the efficiencies of two widely used contracts, one based on sample-average downtime and the other based on cumulative downtime, and identify the supplier's ability to influence the frequency of disruptions as an important factor in determining which contract performs better. We also show that implementing PBC may create high agency cost when equipment is very reliable. This counterintuitive situation arises because the realized downtimes from which the customer might intuit about the supplier's capacity investment are highly uncertain when there are not many samples of downtimes, i.e., when disruptions occur rarely.