Emerged late Quaternary coral reefs show that on the 103 to 105 years time scale the western part of the central New Hebrides arc is divided into several semi-independent uplifted blocks. The blocks are separated by tectonic discontinuities, oriented approximately normal to the arc trend, across which tilt directions or uplift change suddenly. However, none of the blocks is tilted in a direction normal to the arc trend of N20¿W. Two of the discontinuities, across Santo and Malekula Islands, occur near the places where ridges on the north and south flanks of the d'Entrecasteaux fracture zone, a major bathymetric feature on the underthrusting plate, intersect the arc. Each of these tectonic discontinuities approximately coincides with one end of the rupture zone inferred for a shallow thrust-type earthquake sequence in 1965. The discontinuity across Santo also nearly coincides with the south end of the rupture zone inferred for another earthquake sequence in 1973-1974. Uplift of north Malekula, imposed during the 1965 earthquakes, was recorded by emergence of shallow water corals and closely resembles the uplift pattern shown by reef terraces on the 103 to 105 years time scale. Uplift of south Santo that occurred in recent years, possible 1965, also resembles the long-term uplift pattern of that island. These observations indicate that the subducting topography of the d'Entrecasteaux fracture zone has controlled segmentation of the central New Hebrides arc both in terms of seismicity on the 101 years time scale and deformation on the 105 years time scale. Plausible models for buried thrust faults can be made to account for the 1965 uplift pattern. However, there are reasons why a simple elastic dislocation model may be misleading. For example, (1) the uplift data do not adequately constrain the model so that unique fault parameters need be chosen, (2) the block-controlled uplift pattern may interfere with a simple elastic response to faulting, and (3) possible movement on surface faults in north Malekula may have affectaed the displacement pattern. The upper plate in the Santo-Malekula area might best be viewed as a series of loosely coupled blocks whose movements are dominated by underthrusting of rugged topography on the descending plate. Given the Holocene uplift rates for Malekula and the 1965 example of coseismic uplift, it is possible to consider possible recurrence intervals for earthquakes of the same magnitude and uplift as that of 1965. Many assumptions are inherent in this estimate, but it offers valuable perspective. On Malekula, where there was 1.2 m of uplift in 1965, the Holocene uplift rate is 3.5 m/1000 years. Therefore, every 340 years, there must be 1.2 m of uplift to maintain such an average uplift rate. Interseismic recovery of coseismic uplift could shorten considerably the recurrence interval required to maintain this uplift rate. The convergence rate of 11 cm/yr at the New Hebrides arc implies fault slip of 37 m per 340 years. Since this amount of slip is unlikely for the 1965 earthquakes or their predecessors, either some aseismic slip or a seismic recurrence interval of less than 340 years is suspected.