Stress tensor determinations from probable Mio-Pliocene age minor faults along the edge of the Richmond block, northwest of the Alpine fault system of South Island, show pervasive thrusting rather than strike-slip motion as proposed in some analyses. This study, with results almost identical to that of Nicol and Wise (1992) on the other side of the Alpine/Marlborough fault system, shows that the present pattern of regional compression at high angles to the Alpine fault has persisted through much of the Neogene over a width of about 150 km normal to the Alpine fault zone. One explanation might be San Andreas-type models invoking strike-slip motion concentrated in fault zones by highly lubricating gouge coupled with strain partitioning to allow compression in adjacent regions to act at high angles to strike. Such models are difficult to apply to the New Zealand example which now must include deformed forelands on either side of the zone but which shows little or no strike-slip motion. Instead, a model is proposed for a crustal scale ''flower structure'' to accommodate the along-strike convergence of opposite dipping, subducting plates. A regional surface slab comprising Marlborough and its adjacent forelands is undergoing compression normal to the strike of the system while semidetached, deeper plate motion accommodates most of the fundamental, converging, oblique, strike-slip displacements.