The computer model developed for Louisville's Raw Water Master Plan was designed to simulate the operation of the City's system under various drought conditions, patterned after historic droughts. Though the model shows whether the existing system is capable of withstanding the drought(s) occurring during a historic drought, the results do not provide an estimate of the statistical "reliability" of the system. To evaluate reliability, statistical analyses of the historic streamflow record were conducted to characterize past droughts in terms of magnitude, duration, and severity. The drought analyses revealed that giving any period a specific drought label (e.g., the 50-year RI, 24-month duration drought) is not a clear-cut decision and can be misleading. Any such drought could also be shown to be comprised of a multitude of both dependent and independent droughts of shorter durations and both lesser and greater severity. Rather than selecting the "worst drought of record" as the design period for the Master Plan, drought periods in the historical record were located on the MFD curves to identify candidate simulation periods for modeling the City's system at acceptable levels of protection, defined by both duration and frequency of occurrence.