We investigated whether observed changes in seismicity rate following the 1992 Landers and 1995 Kobe earthquakes correspond with predictions from Coulomb stress transfer models. To quantify the relationship between Coulomb stress change and seismicity rate change, we compared spatial changes in observed seismicity rates to the modeled changes in Coulomb failure stress (ΔCFS) for the two subject earthquakes. We begin with the simplest and most common ΔCFS model, which assumes that regional stress orientations are uniform and the stress change is resolved on optimally oriented strike-slip planes. We then extended the ΔCFS models to include the more geophysically realistic assumptions of spatially varying regional stress orientations and optimally oriented planes in three dimensions. For all the tested ΔCFS models, we found that virtually everywhere seismicity rate changes could be resolved, the rate increased regardless of whether the ΔCFS theoretically promoted or inhibited failure and that we were not able to reliably resolve seismicity rate decreases. An improvement in the correlation between the sign of the ΔCFS and where rate increases occur when estimating the ΔCFS on optimally oriented planes in three dimensions rather than two dimensions is a consequence of using optimally oriented planes and the ability to primarily observe rate increases. The region of positive ΔCFS increases by allowing more degrees of freedom so the correlation with positive rate change appears to increase. The areas of highest seismicity rate following the main shock did not correlate with the sign or the magnitude of the ΔCFS. Rather, the areas most active after the main shock were also the most active before the main shock.