The origin of the multiple concentric rings that characterize lunar impact basins, and the probable depth and diameter of the transient crater have been widely debated. As an alternative to prevailing "megaterrace" hypotheses, we propose that the outer scarps or mountain rings that delineate the topographic rims of basins-the Cordilleran at Orientale, the Apennine at Imbrium, and the Altai at Nectaris-define the transient cavities, enlarged relatively little by slumping, and thus are analogous to the rim crests of craters like Copernicus; inner rings are uplifted rims of craters nested within the transient cavity. The magnitude of slumping that occurs on all scarps is insufficient to produce major inner rings from the outer. These conclusions are based largely on the observed gradational sequence in lunar central uplifts:. from simple peaks through somewhat annular clusters of peaks, peak and ring combinations and double ring basins, culminating in multiring structures that may also include peaks. In contrast, belts of slump terraces are not gradational with inner rings. Terrestrial analogs suggest two possible mechanisms for producing rings. In some cases, peaks may expand into rings as material is ejected from their cores, as apparently occurred at Gosses Bluff, Australia. A second process, differential excavation of lithologically diverse layers, has produced nested experimental craters and is, we suspect, instrumental in the formation of terrestrial ringed impact craters. Peak expansion could produce double-ring structures in homogeneous materials, but differential excavation is probably required to produce multiring and peak-in-ring configurations in large lunar impact structures. Our interpretation of the representative lunar multiring basin Orientale is consistent with formation of three rings in three layers detected seismically in part of the Moon-the Cordillera (basin-bounding) ring in the upper crust, the composite Montes Rook ring in the underlying, more coherent "heald" crust, and an innermost, 320-km ring at the crust-mantle interface. Depth-diameter ratios of 1 10to 1 15 are consistent with this interpretation and suggest that volumes of transient cavities and hence of basin ejecta may be considerably greater than commonly assumed. ?? 1978.