Global Navigational Satellite Systems (GNSS) are being employed to augment seismic instrumentation to record large, dynamic displacements and accelerations from large earthquakes. To date, however, there have been only a few tests that independently characterize the GNSS at frequencies and displacements that occur during large earthquakes (a number of error sources might influence such GPS result, including loss of lock or bias in signal tracking loops). Many of these tests consist of replaying the observed accelerations for select earthquakes recorded by seismic instruments through a shake-table on which a GNSS antenna is attached. Then the derived displacement from the accelerometer is compared with the displacement estimated from the GNSS system, or the GNSS derived acceleration is compared with the acceleration of the shake table. Neither comparison is optimal since derived quantities are used, and in particular, displacements derived from acceleration data have many sources of error at long periods. Another approach is to test the response of the GNSS receiver using a GNSS-simulator where synthetic GNSS signals are generated that mimic the signals that are actually received. Ebinuma and Kato (Earth Planet Space, 2012) describe a series of controlled tests using this approach with three different GNSS receivers. As a 'real world' test, we performed similar experiments using a shake table, in open air with normal views of GNSS satellites, with controlled displacement inputs but, importantly, measured the displacement and acceleration of this table independently. We used a single-axis shake-table having up to 40 cm horizontal displacement and independently measured the position of the stage to better than 0.1 mm (from table servo loop optical reference; accelerations measured by accelerometers attached to moving part of stage). We tested five different GNSS receivers recording both GPS and GLONASS at 50 samples per second (sps), with the exception of the Trimble NetRS, which is widely used in the Plate Boundary Observatory, that recorded 10 sps GPS data only. To determine the system response, we tested each receiver using both a rounded Heaviside (step) function (40, 10, and 2 cm) and a variety of sinusoid signals from 0.2 to 20 Hz spanning a range in accelerations between 0.02 to 3.0 g. The results of these real-world tests will be compared with similar measurements by Ebinuma and Kato (2012), who used a GPS simulator.