Inverse time overcurrent relays are widely used for the protection of distribution networks. They detect the current caused by a short circuit fault and after an appropriate time delay disconnect the faulted item of plant from the network. Many electrical faults, especially those occurring on underground cables, are intermittent in nature and are commonly referred to as “pecking” faults. Such faults are highly unstable, and the fault arc repeatedly flashes over and then recovers. This “on-off” behaviour of the fault current adversely affects the operating times of overcurrent relays; and this can create grading problems, especially if a feeder is protected with a mixture of electro-mechanical, static and numerical relays. For example if the amplitude of the fault current changes from “high-low-high… etc” an upstream relay may eventually initiate the spurious tripping of a circuit breaker and black-out a large section of the network, that includes healthy feeders and the downstream faulted component. The actual response of a relay to a pecking fault depends on the reset behaviour of the actual or virtual operating disc. For example, if the duration of one fault current pulse is insufficient to operate the relay and if the reset time of the relay is shorter than the time between successive pulses of fault current, then the relay will never operate; however, if the reset time is longer than the off current period, the disc will effectively integrate the effect of the fault current pulses and after multiple pulses trip the relay. This paper investigates the impact of repetitive pecking faults on the operating times of different generations of overcurrent relays and recommends a solution to the coordination problem seen on a network during an intermittent fault.