Modern Lightning Detection and Warning Systems (LDWSs) provide the technology needed to support prompt and effective alerts for the protection of personnel and facilities. However, the effectiveness of such warnings depends critically on the actual procedures used to curtail operations and effect evacuations to safe locations. In this paper we show how probabilistic risk assessment can be used to develop lightning hazard control procedures to obtain an optimal risk/benefit ratio. The approach uses Monte Carlo simulation to model the random behavior of storms at a particular location. The model also considers existing or proposed procedures used to declare warning, alarm, and all-clear conditions. Variations in risk and the costs associated with personnel sheltering are calculated as direct outputs of the analysis. We demonstrate the application of this approach using the LDWS at Los Alamos National Laboratory for outdoor high-explosive operations at a major test facility.