It has been proposed that the crustal stress field contains small-length-scale heterogeneity of much larger amplitude than the uniform background stress. This model predicts that earthquake focal mechanisms should reflect the loading stress rather than the uniform background stress. So, if the heterogeneous stress hypothesis is correct, focal mechanisms before and after a large earthquake should align with the tectonic loading and the earthquake-induced static stress perturbation, respectively. However, I show that the off-fault triggered aftershocks of the 1992 M7.3 Landers, California, earthquake align with the same stress field as the pre-Landers mechanisms. The aftershocks occurred on faults that were well oriented for failure in the pre-Landers stress field and then loaded by the Landers-induced static stress change. Aftershocks in regions experiencing a 0.05 to 5 MPa coseismic differential stress change align with the modeled Landers-induced static stress change, implying that they were triggered by the stress perturbation. Contrary to the heterogeneous stress hypothesis, these triggered aftershocks are also well aligned with the pre-Landers stress field obtained from inverting the pre-Landers focal mechanisms. Therefore, the inverted pre-Landers stress must represent the persistent background stress field. Earthquake focal mechanisms provide an unbiased sample of the spatially coherent background stress field, which is large relative to any small-scale stress heterogeneity. The counterexample provided by the Landers earthquake is strong evidence that the heterogeneous stress model is not widely applicable.