The NASA/NGA Shuttle Radar Topography Mission (SRTM) collected interferometric radar data to generate a near-global topography dataset for latitudes below 60°. The primary goal was to produce a dataset with globally consistent and quantified errors. Extensive ground campaigns were conducted to collect ground truth data for validation. This paper documents the results of the SRTM validation effort, focusing on characterizing the height and position errors of the dataset. The SRTM used interferometric SAR technology to measure distances between known points and the Earth's surface, allowing for the triangulation of surface locations and inference of surface height. The mission aimed to produce DEMs with globally consistent and known errors. The paper reviews the ground-truth data used and summarizes the characteristics of residual errors after calibration and continental least squares error adjustments. The verification of the SRTM DEMs involved extensive ground control points (GCPs) collected using kinematic Global Positioning System (KGPS) transects. The GCPs were processed to yield nearly 9.4 million samples covering six continents with an accuracy of approximately 50 cm. The data were divided into two sets: one for validating absolute height accuracy and another for estimating geolocation accuracy. The error sources for interferometric SARs are well understood, including static and time-varying errors. Static errors, such as baseline roll errors, can be calibrated using natural or man-made targets, while time-varying errors, like phase errors, are partially compensated through dynamic calibration and least squares adjustments. The paper examines the spatial characteristics of the errors, including long wavelength errors and random errors. Long wavelength errors, due to residual motion errors, have a magnitude of about 2 meters and are uncorrelated with the random errors. Random errors, caused by instrument thermal noise and geometric decorrelation effects, have a very short spatial correlation length (45-90 meters) and exhibit geographical variability. The SRTM's performance is assessed using various globally distributed datasets, showing that the absolute height error exceeds the mission goal of 16 meters (90% error) by almost a factor of two. The continent-wide absolute height error is estimated by subtracting the DEM height from the GPS ground-truth after interpolation. The geolocation error was estimated using image and height matching of the KGPS transects. The paper concludes with a summary of the SRTM performance, highlighting the continent-wide absolute height error and geolocation error, and providing estimates of the combined error characteristics on a cell and sub-cell basis.The NASA/NGA Shuttle Radar Topography Mission (SRTM) collected interferometric radar data to generate a near-global topography dataset for latitudes below 60°. The primary goal was to produce a dataset with globally consistent and quantified errors. Extensive ground campaigns were conducted to collect ground truth data for validation. This paper documents the results of the SRTM validation effort, focusing on characterizing the height and position errors of the dataset. The SRTM used interferometric SAR technology to measure distances between known points and the Earth's surface, allowing for the triangulation of surface locations and inference of surface height. The mission aimed to produce DEMs with globally consistent and known errors. The paper reviews the ground-truth data used and summarizes the characteristics of residual errors after calibration and continental least squares error adjustments. The verification of the SRTM DEMs involved extensive ground control points (GCPs) collected using kinematic Global Positioning System (KGPS) transects. The GCPs were processed to yield nearly 9.4 million samples covering six continents with an accuracy of approximately 50 cm. The data were divided into two sets: one for validating absolute height accuracy and another for estimating geolocation accuracy. The error sources for interferometric SARs are well understood, including static and time-varying errors. Static errors, such as baseline roll errors, can be calibrated using natural or man-made targets, while time-varying errors, like phase errors, are partially compensated through dynamic calibration and least squares adjustments. The paper examines the spatial characteristics of the errors, including long wavelength errors and random errors. Long wavelength errors, due to residual motion errors, have a magnitude of about 2 meters and are uncorrelated with the random errors. Random errors, caused by instrument thermal noise and geometric decorrelation effects, have a very short spatial correlation length (45-90 meters) and exhibit geographical variability. The SRTM's performance is assessed using various globally distributed datasets, showing that the absolute height error exceeds the mission goal of 16 meters (90% error) by almost a factor of two. The continent-wide absolute height error is estimated by subtracting the DEM height from the GPS ground-truth after interpolation. The geolocation error was estimated using image and height matching of the KGPS transects. The paper concludes with a summary of the SRTM performance, highlighting the continent-wide absolute height error and geolocation error, and providing estimates of the combined error characteristics on a cell and sub-cell basis.