At subduction zones where convergence is not perpendicular to the trench (i.e., oblique), slip vectors of interplate thrust earthquakes are often systematically deflected away from the expected plate vector, probably by deformation of the leading edge of the upper plate. Near the epicenters of great thrust earthquakes of Mw≥8.0 in this century that occurred at trenches, systematic deflections of slip vectors for recent (1977--1992.5) interplate thrust earthquakes are significantly less than global averages. Statistical tests show that great earthquakes nucleate where deviations of slip vectors (residuals) are smaller and plate convergence is faster but not preferentially where the angle of obliquity is smaller than global averages. The small slip vector residuals suggest that great earthquakes nucleate where forces in the forearc due to plate convergence are sustained elastically, rather than anelastically, and that shear stress on subduction thrust faults with great earthquakes is not necessarily higher than on faults without them. Shear stress magnitude is less important than rheology in determining the seismic behavior of subduction zones. An observed decrease in the slip vector residuals with increasing rates of trench-normal convergence is interpreted as evidence that the forearc is increasingly cooled and strengthened by faster subduction. For the forearc to cool with increasing convergence rate, the shear stress on the subduction thrust fault must be less than about 40 MPa. Lower temperatures on the fault surface and the possibility that thrust faults below undeforming forearcs are smoother than those below deforming forearcs both act to increase the frictional instability of the fault zone. Great thrust earthquakes at subduction zones seem to occur where convergence is fast and the forearc is cool and predominantly elastic. The ability of the forearc of the upper plate to store elastic strain energy appears to be a major factor controlling where great subduction thrust earthquakes can and cannot occur. ¿ American Geophysical Union 1993