Over the last decade, the 1-sample-per-second kinematic Global Positioning System (GPS) has been used as a displacement sensor in earthquake observations and for structural health monitoring. Many researchers in both seismology and engineering have expressed the desire for higher-sample-rate (10-sample-per-second or higher) GPS data to acquire high-frequency displacement information. We performed several shake table tests of GPS observation on 29 April, 2009 for the purpose of evaluating the performance of high-rate kinematic GPS. We found that the accuracy of high-rate kinematic GPS depended on antenna movement, but was independent of receiver sampling rate. The errors in kinematic GPS measurements during the periods of strong shaking were systematically larger than those during the static periods. Furthermore, we found that these large errors were coincident with large accelerations and jerks in the motions experienced by the GPS receivers and antennas. Observations from the 2010 earthquake in Maule, Chile (M 8.8) indicated that strong ground motions can degrade the accuracy of high-rate kinematic GPS measurements. Significant jerks and/or accelerations can cause GPS units to temporarily lose tracking on satellite signals and lead to gaps in GPS-recorded seismograms.