Focal depths from over 1000 earthquakes occuring between 1977 and 1983 in the Imperial Valley-southern Peninsular Ranges are used to study relationships between the depth of seismicity, heat flow, and crustal structure. This study used relocated A and B quality events from the California Institute of Technology catalog that were carefully selected to insure focal depth precision of ¿2 km. Regional variations in focal depth appear to be related to regional heat flow variation, whereas local variations in focal depth, especially in the central Imperial Valley, may be related to crustal structure. These variations are studied by rheologic modeling. A comparison of focal depths of earthquakes occurring before and after the October 15, 1979 (M=6.6), earthquake indicates that aftershocks during the first 2 months of the sequence were 2--3 km deeper than earthquakes occurring in other time periods. The deepest earthquakes in Imperial Valley are spatially associated with a subbasement dome near the northern end of the Imperial fault. This dome coincides with the region where the Imperial fault undergoes a transition from stick-slip behavior of aseismic fault creep. Models of slip during the 1979 mainshock are also compared with premainshock and postmainshock seismicity. A relocation of the 1940 (M=7.1) mainshock suggests that this sequence began by rupturing the same portion of the fault that experienced maximum slip during the 1979 mainshock.