We present results of a series of numerical experiments designed to test hypothetical mechanisms that drive deformation in the New Madrid seismic zone. Experiments are constrained by subtle topography and the distribution of seismicity in the region. We use a new boundary element algorithm that permits calculation of the three-dimensional deformation field. Surface displacement fields are calculated for the New Madrid zone under both far-field (plate tectonic scale) and locally derived driving strains. Results demonstrate that surface displacement fields cannot distinguish between either a far-field simple or pure shear strain field or one that involves a deep shear zone beneath the upper crustal faults. Thus, neither geomorphic nor geodetic studies alone are expected to reveal the ultimate driving mechanism behind the present-day deformation. We have also tested hypotheses about strain accommodation within the New Madrid contractional step-over by including linking faults, two southwest dipping and one vertical, recently inferred from microearthquake data. Only those models with step-over faults are able to predict the observed topography. Surface displacement fields for long-term, relaxed deformation predict the distribution of uplift and subsidence in the contractional step-over remarkably well. Generation of these displacement fields appear to require slip on both the two northeast trending vertical faults and the two dipping faults in the step-over region, with very minor displacements occurring during the interseismic period when the northeast trending vertical faults are locked. These models suggest that the gently dipping central step-over fault is a reverse fault and that the steeper fault, extending to the southeast of the step-over, acts as a normal fault over the long term. ¿ American Geophysical Union 1994