GPS users need to know about the accuracy of the position and velocity information produced by the technique. The positioning accuracy of the technique depends mainly on observing session duration. The accuracy of GPS was previously studied using samples of 10-30 consecutive days. Namely, the effect of seasons was not taken into consideration. For instance, an annual cycle plays crucial role on GPS vertical rates. In order to quantify the global disturbance on positioning due to seasonal differences, here we designed a GPS experiment using the stations of International GNSS Service (IGS). GPS data were sampled from a two-year range not from several days. The data were processed using GIPSY/OASIS II. The results show that the RMS is greater at the equatorial regions. The new prediction of the accuracy becomes poorer especially for 8-24 h sessions. In addition, the traditional accuracy formulation needs to be expanded including one additional term. (C) 2019 Elsevier Ltd. All rights reserved.