NASA's Ice, Cloud, and land Elevation Satellite (ICESat), which operated between 2003 and 2009, made the first satellite-based global lidar measurement of Earth's ice sheet elevations, sea-ice thickness and vegetation canopy structure. The primary instrument on ICESat was the Geoscience Laser Altimeter System (GLAS), which measured the distance from the spacecraft to Earth's surface via the roundtrip travel time of individual laser pulses. GLAS utilized pulsed lasers and a direct detection receiver consisting of a silicon avalanche photodiode (Si APD) and a waveform digitizer. Early in the mission, the peak power of the received signal from snow and ice surfaces was found to span a wider dynamic range than planned, often exceeding the linear dynamic range of the GLAS 1064-nm detector assembly. The resulting saturation of the receiver distorted the recorded signal and resulted in range biases as large as ~50 cm for ice and snow-covered surfaces. We developed a correction for this "saturation range bias" based on laboratory tests using a spare flight detector, and refined the correction by comparing GLAS elevation estimates to those derived from Global Positioning System (GPS) surveys over the calibration site at the salar de Uyuni, Bolivia. Applying the saturation correction largely eliminated the range bias due to receiver saturation for affected ICESat measurements over Uyuni and significantly reduced the discrepancies at orbit crossovers located on flat regions of the Antarctic ice sheet.