High-resolution cluster analysis (multiple-event relocation) of earthquakes and other seismic sources is developed as a tool for assembling catalogs of reference events, especially those whose locations can be determined with an accuracy of 5 km or better [Ground Truth (GT) 5]. We use the Hypocentroidal Decomposition (HDC) method of Jordan and Sverdrup (1981), which is well suited to the rigorous statistical analysis required for this task. Candidate reference events typically arise from local seismic networks and from temporary deployments for aftershock studies that can yield very high-resolution hypocenters that, nevertheless, must be validated. We utilize arrival time data (as reported to the International Seismological Centre and to the U.S. Geological Survey's National Earthquake Information Center) at regional and teleseismic distances in the cluster analysis to validate candidate reference events, and in some cases, to generate new reference events. HDC analyses have now been performed on a number of earthquake and explosion sequences in Eurasia and Africa, resulting in reference events with locations known to GT5 accuracy. In this paper we review and evaluate our analyses of these clusters to date, and address problem areas. In particular, we find that some candidate reference events cannot be validated because either the reported local network solutions are in error, or the coverage of reported arrival times used in the HDC analysis is not sufficient to constrain the locations. Some discrepancies may arise when local networks locate small precursors or low-energy early stages of rupture in larger earthquakes, while teleseismic stations record only the main pulse of energy release. We have found several cases in which there appear to be systematic biases in the time base used for local network solutions. In another case, we obtained reference event locations from two different sources for the same cluster.