Range measurements from the lidar instrument carried aboard the Clementine spacecraft have been used to produce an accurate global topographic model of the Moon. This paper discusses the function of the lidar; the acquisition, processing, and filtering of observations to produce a global topographic model; and the determination of parameters that define the fundamental shape of the Moon. Our topographic model; a 72nd degree and order spherical harmonic expansion of lunar radii, is designated Goddard Lunar Topography Model 2 (GLTM 2). This topographic field has an absolute vertical accuracy of approximately 100 m and a spatial resolution of 2.5°. The field shows that the Moon can be described as a sphere with maximum positive and negative deviations of ∼8 km, both occurring on the farside, in the areas of the Korolev and South Pole‐Aitken (S.P.‐Aitken) basins. The amplitude spectrum of the topography shows more power at longer wavelengths as compared to previous models, owing to more complete sampling of the surface, particularly the farside. A comparison of elevations derived from the Clementine lidar to control point elevations from the Apollo laser altimeters indicates that measured relative topographic heights generally agree to within ∼200 m over the maria. While the major axis of the lunar gravity field is aligned in the Earth‐Moon direction, the major axis of topography is displaced from this line by approximately 10° to the east and intersects the farside 24° north of the equator. The magnitude of impact basin topography is greater than the lunar flattening (∼2 km) and equatorial ellipticity (∼800 m), which imposes a significant challenge to interpreting the lunar figure. The floors of mare basins are shown to lie close to an equipotential surface, while the floors of unflooded large basins, except for S.P.‐Aitken, lie above this equipotential. The radii of basin floors are thus consistent with a hydrostatic mechanism for the absence of significant farside maria except for S.P.‐Aitken, whose depth and lack of mare require significant internal compositional and/or thermal heterogeneity. A macroscale surface roughness map shows that roughness at length scales of 10 1 –10 2 km correlates with elevation and surface age.