Fault slip rate is one of the most important subjects in active tectonics research, which reveals the activity and seismic potential of a fault. Due to the improvement of dating precision with the development of dating methods, Holocene geological markers, even the young markers of thousands or hundreds of years old, are widely used in fault slip rate calculation. Usually, uncertainties from a single event and erosion of the accumulated offsets are involved in fault slip rate determination. Two types of uncertainties are related to a single event; the first is the time elapsed since the latest (the most recent) event; the second is the period since the formation of the geological marker to the occurrence of the first event. High-slip-rate faults are more sensitive to these uncertainties than low-slip-rate faults. In this study, we studied quantitatively the effects of a single event on fault slip rate following the three classic earthquake models: the characteristic earthquake, uniform slip and variable slip models. We suggest that the erosion of the accumulated offset–lateral erosion on a strike-slip fault, should also be considered in fault slip estimation. Therefore, we propose a differential method to obtain a reliable fault slip rate. In the differential method, the slip rate is the ratio of offset differentials and corresponding age differentials between the older and younger terraces along strike-slip faults. This kind of differential method could avoid the uncertainties from the first and latest events, as well as that from the lateral erosion. By applying the differential method, we got the revised slip rates of ∼5–10 mm/year on the Altyn Tagh and Kunlun faults. These low slip rates could fit previous geodetic and geological fault slip rates and shortening rates as well as the millennial recurrence intervals of strong earthquakes along the major segments of these faults.