Abstract Using an automated “aspect ratio method” to evaluate shear-wave anisotropy from earthquakes recorded on arrays of three-component digital seismographs over the Bucaramanga nest of Colombia, the East Rift Zone of Hawaii, and the Phlegraean Fields uplift area of Italy, a shear-wave velocity anisotropy of 4–7% is observed. The anisotropy is attributed to tectonic processes acting either to align anisotropic minerals or to align stress-related cracks. At most fixed seismographs, the polarization of the leading S-phase is consistent from event to event, independent of earthquake location. On a regional basis, the polarization may vary from station to station, but a regional trend is apparent. For events from the Bucaramanga nest, Colombia, a small seismogenic volume at 160 km depth in a subducted plate, shear-wave splitting averages 0.5 ±0.1 s for more than 100 events. The polarization direction of the fast shear wave is 355° N± 15°, parallel to the strike of the Benioff zone. At the Phlegraean Fields, in Italy, which represent an active Quaternary caldera, the average shear-wave splitting observed from 51 events is 0.2 s, corresponding to a minimum velocity anisotropy of 7%. The strike of the inferred maximum compressive stress is consistent with values obtained from fault-plane solutions. In Hawaii, the polarization of the leading shear wave tends to parallel the strike of the East Rift Zone. Average velocity anisotropy is about 5% if one assumes that the entire travel path contributes to the observed splitting. However, station variations in fast shear-wave polarization direction imply that some of the splitting is due to near-station effects.