This research deals with the nature of the rupture process of large earthquakes whose low-frequency behaviors have been found to be anomalous. The investigation will use teleseismic data from high-performance global seismic networks. The approach will combine into a single inversion scheme the information from two complementary procedures: a time-domain methodology for estimating the spatial locus of moment release as a function of time, and a spectral methodology for estimating low-degree polynomial moments. The first uses broadband body waves to get high spatial and temporal resolution, while the latter uses low-frequency surface waves and free oscillations to place precise integral constraints on the entire source process. The specific objectives of the project include the study of the source complexity associated with three classes of anomalous events: slow earthquakes on the oceanic ridge-transform system, intermediate-focus earthquakes with large CLVD (volume change) components, and complex earthquakes of long duration in continental crust. This research is a component of the National Earthquake Hazard Reduction Program.