The disclosed methods and systems prioritize network devices and components that more likely result in a network outage by evaluating their current and historical outage information relative to a dynamically-generated, statistical threshold based on such outage information. The prioritized list of network devices can be organized by, for example, quantity of outages, number of outage minutes, and/or number of customer-outage minutes so as to provide diagnosis sequences by type of network device, network location, type of component within the network device, and/or by operation during particular time periods/events.