During times of thunderstorm activity, the power utility serving metropolitan New York enters a potentially costly “thunderstorm watch” mode of operation which is designed to prevent a major power outage caused by lightning. To evaluate the usefulness of real-time lightning-location data in better defining required watch periods, the recorded log of actual watches is compared to a derived log of watch periods based solely on archived cloud-to-ground lightning data from the SUNY–Albany Lightning Detection Network. The analysis period spans February–July of 1984. Simple objective criteria define the initiation and termination times of network-derived watches. A lightning derived watch begins when two lightning flashes occur within 5 min of each other anywhere within a 20-km extension of the borders of the utility's operating region; a derived watch ends when no lightning is observed within the same region for a period of at least 15 min. Of 36 watches common to both databases, the lightning derived initiation and termination times precede the actual logged times by an average of 37 and 66 min, respectively, reducing watch durations by about 0.5 h, on average. Cases where the derived watches occur without corresponding logged alerts, and vice versa, are investigated and interpreted in light of the available meteorological data. We conclude that real-time access to lightning location data would better allow the power utility to deal with the cloud-to-ground lightning threat in two important respects: 1) more timely warnings, and 2) reduced periods of costly “thunderstorm watch” operation.