The aftershock distribution of the 1994 Arthur's Pass earthquake, Mw6.7, is unusual for a reverse faulting event in that it extends 12 km NNW and 30 km SSE of the actual fault plane, which strikes NE-SW. We have used several methods to infer the regional stress field in the region, including geodetic results, earthquake mechanisms, and inversion of P wave polarity data for the stress tensor orientation. The inversion method is new and does not require the focal mechanisms of the events used. It also incorporates the Coulomb failure criterion. All results point to a stress field favoring strike-slip faulting, not thrusting, with near-horizontal &sgr;1 and &sgr;3 principal axes striking at 298¿ and 28¿. Using dislocation theory, we calculate the stress induced by the Arthur's Pass earthquake and its largest aftershock (a strike-slip event) and add this to the regional field. There is a fair correspondence between the hypocenters of aftershocks away from the mainshock fault plane and regions of high induced Coulomb Failure Stress (CFS) on optimally oriented fault planes. However, there are regions of high induced CFS that are devoid of aftershocks. It appears that earthquake slip in this region of oblique (19¿) plate convergence is, as observed elsewhere, partitioned into components parallel and perpendicular to the plate margin. Most of the slip is parallel, as occurs on the nearby dextral Alpine fault, the boundary between the Pacific and Australian plates. However, occasional reverse events, such as the Arthur's Pass earthquake, account for at least some of the perpendicular component of slip and the uplift that produced the Southern Alps.