In nuclear power plant siting, redetermination of hypocenters of instrumentally recorded earthquakes in the site region may be better than relying on routinely determined hypocenter locations listed in standard earthquake catalogues. Routinely determined hypocenters, particularly those of small earthquakes or of earthquakes occurring before the mid-1950s, may be so mislocated that they present a misleading picture of the seismo-tectonics of the site region. The preferred method for redetermining a hypocenter depends on the earthquake being considered. The hypocenters of many pre-1954 earthquakes may be more accurately redetermined by now-standard computerized location techniques that were not available when the earthquakes were initially catalogued. Some important earthquakes are so poorly recorded that they cannot be assigned a hypocenter that is mathematically unique; however, it may be possible to pose a hypothesis such that the significance of the earthquake to a proposed power plant can be tested without explicitly locating the earthquake. A well-recorded earthquake whose hypocenter is suspected of being biased by lateral variations of seismic wave velocities can be more accurately located by ray-tracing methods, provided that the lateral variations of velocity are known. In an attempt to infer a fault or seismic zone from a group of hypocenters, a relative-location method—such as the master-event method or joint hypocenter determination—should be used. All methods for determining hypocenters involve assumptions over which seismologists may disagree; in the safety analysis of a nuclear power plant, these assumptions must be explicitly discussed and shown to be conservative as to their implication for the design of the plant. In addition, confidence ellipses should be given for the hypo-central coordinates, in order to assess the errors that might occur even if the assumptions of the location method are correct.