Analysis of the sensitivity of laser ranging errors to surface conditions indicates that predicted single-shot range errors are primarily dependent on surface slope. Range errors are less sensitive to variations in surface roughness or reflectivity. Values of total surface slope and roughness for nine terrestrial landforms, derived from digital elevation data at a 186 m length scale, vary from 2/spl deg/ to 40/spl deg/ and 0.8 to 15 m, respectively, at a 90% frequency of occurrence. This range of surface morphologies yields a variation in single shot laser ranging error from 0.4 to 8 m, assuming system parameters for the proposed Topographic Mapping Laser Altimeter (TMLA) and a nominal 30% surface reflectivity. The total elevation accuracy of data obtained via satellite laser altimetry, although dominated by the range error, is also a function of additional error sources, including orbit ephemeris, atmospheric, and calibration errors. Averaging of multiple laser measurements improves the vertical accuracy of the elevation data by statistical reduction of random errors. During a three-year mission, two to three laser measurements will be acquired, on average, for each 200-m footprint at low to moderate latitudes, accounting for the latitudinal variation of ground track spacing and cloud cover. For high-latitude regions, the narrow spacing of satellite ground tracks in a polar orbit will provide frequent repeat observations yielding, on average, 4 to 25 measurements of each footprint over the Antarctic and Greenland ice sheets. Averaging of these multiple repeat observations at high latitude will yield an improvement in vertical accuracy by a factor of two to five. >