An important issue in determining the accuracy of global positioning system (GPS) satellite ionospheric measurements is the instrumental delay biases between the L-band frequencies in both the satellites and the receivers. These differential L1-L2 biases must be measured and removed from the GPS measurements before an accurate estimate of the total electron content can be obtained. The results from the measurements indicate that the day-to-day variations of the satellite differential biases are quite well over a five-week time span, with a variation of less than 0.3-ns differential delay (one sigma). A follow-up experiment conducted two years later showed that the satellite biases had not changed significantly over this longer time span. When the prelaunch calibration values are compared with the experimental bias estimates, two of the four satellite pairs show excellent agreement and two differ significantly, indicating that prelaunch calibrations should be used with caution. >