The main source of error in the estimation of TEC (total electron content) from dual Global Positioning System (GPS) data is the effect of the differential satellite and receiver instrumental delay biases. These biases are normally estimated simultaneously with the TEC. However, the additional estimation of the instrumental biases may constitute an insurmountable burden in some practical applications like real-time estimation of TEC, or the estimation may be difficult or correlated to the ionospheric parameters, particularly in situations where the TEC behavior may be harder to model (equatorial or auroral zone, ionospheric storms, etc.). A priori values of the instrumental biases, estimated under good conditions or with global networks, could solve those problems if we could determine how stable those instrumental biases are in time and how often we need to check or reestimate their values. In this paper we will present our estimation of the GPS satellite and receiver instrumental biases from 19 months of data and the study of their variation during that time. We will also show some situations of changes in the instrumental biases and the possible influence of antispoofing (AS). The main conclusion of this work is that the variation of the estimated differential GPS satellite biases during the 19 months is smaller than 1 ns (1 ns=2.86¿1016 e/m2) in most of the cases, with a mean RMS of 0.15 ns. For the GPS receivers used, that variation is greater than for the satellites, with the larger variations corresponding to physical changes in the receivers. The difference of the estimated differential instrumental biases between two consecutive days is in practically all cases smaller than 0.5 ns for the GPS satellites and smaller than 1 ns for the GPS receivers. Regarding the influence of AS, we have detected some significant changes in the instrumental biases of some satellites and some stations whether AS is activated or not. Our main conclusion is that due to the stability of the GPS instrumental biases, only an estimation or calibration of them (under optimal conditions) from time to time is required.¿ 1997 American Geophysical Union