We use four geodetic satellite systems (Global Positioning System [GPS], European Remote Sensing [ERS], RADARSAT, and Satellite Pour l'Observation de la Terre [SPOT]) to measure the permanent deformation field produced by the İzmit earthquake of 17 August 1999. We emphasize measurements from interferometric analysis of synthetic aperture radar (SAR) images acquired by ERS and RADARSAT and their geodetic uncertainties. The primary seismological use of these data is to determine earthquake source parameters, such as the distribution of slip and the fault geometry. After accounting for one month's postseismic deformation, tropospheric delay, and orbital gradients, we use these data to estimate the distribution of slip at the time of the İzmit mainshock. The different data sets resolve different aspects of the distribution of slip at depth. Although these estimates agree to first order with those derived from surface faulting, teleseismic recordings, and strong motion, careful comparison reveals differences of 40% in seismic moment. We assume smooth parameterization for the fault geometry and a standard elastic dislocation model. The root mean square residual scatter is 25 mm and 11 mm for the ERS and RADARSAT range changes, respectively. Our estimate of the moment from a joint inversion of the four geodetic data sets is M0 = 1.84 × 1020 N m, a moment magnitude of Mw 7.50. These values are lower than other estimates using more realistic layered earth models. Given the differences between the various models, we conclude that the real errors in the estimated slip distributions are at the level of 1 m. The prudent geophysical conclusion is that coseismic slip during the İzmit earthquake tapers gradually from approximately 2 m under the Hersek delta to 1 m at a point 10 km west of it. We infer that the Yalova segment west of the Hersek delta may remain capable of significant slip in a future earthquake.