The observed differences between complex craters and multi-ringed basins are described and a combined “transient cavity/displaced zone/melt cavity” model is explored to account for the observations. In this “nested melt-cavity model”, the increasing influence of the percentage of the target undergoing impact melting at the sub-impact point with increasing size (the differential melt scaling of Cintala and Grieve ) causes fundamental changes in the nature of the transient cavity, its relationship to the displaced zone, and its short-term collapse behavior. The transition from complex craters to two-ring basins involves expansion of the melting front into the displaced zone, formation of a two-component excavation cavity, and a concomitant radial expansion outward of the zone of maximum elastic rebound. The short-term modification stage is then dominated by the strength differences between the fluid melt in the inner cavity and rocks of the displaced zone; the highly shocked rocks at the outer margin of the expanded parabolic melting front rebound to form the expanded peak ring, moving upward and laterally inward, easily displacing fluid melt filling the inner depression. At larger sizes, differential melt scaling causes the peak ring diameter to expand relatively more rapidly than the basin rim, and upon collapse, the increased volume of melt ponds in the rebounded melt cavity inside the expanding peak ring. As the transient melt cavity further increases proportionally in size, it penetrates through the base of the displaced zone with significant consequences. The resulting modification stage now incorporates into the collapse process inward and upward movement along the base of the displaced zone; listric failure occurs inward into the fluid melt cavity beginning at the edge of the melt cavity and extending out along the base of the displaced zone up to the base of the rim structural uplift (at ∼1.5 crater radii). This forms an additional outer ring and a resulting megaterrace, modifying the radial ejecta on the collapsed rim to form a domical facies. At multi-ring basin scales, the significantly deeper penetration that occurs in the expanding melt cavity accounts for the maximum crustal thickness decrease that occurs inside the peak ring in the final basin. Early onset and higher density of peak ring basins on Mercury is predicted by higher mean impact velocity and differential melt scaling.