Abstract The wealth of Sentinel‐1 data in eastern Tibet allows to discuss whether InSAR‐derived velocity maps measure ground motion in the ITRF reference frame with sufficient accuracy for large‐scale tectonic applications. High elevation, moderate relief and low vegetation cover in eastern Tibet ensure reliable multi‐temporal InSAR results. We automatically process seven 1,200 km long ascending and descending orbits, divided into 2 or 3 overlapping segments along‐track. Linear phase ramps in range and azimuth are removed from interferograms. The ramps and the flattened interferograms are separately inverted into time series and analyzed. The ramps are modeled by their Solid Earth Tides (SET) contribution, seasonal variations, and a trend in time. SET oscillations are clearly visible in the ramp time series. Seasonal tidal amplifications of probable atmospheric origin are also evidenced. Ramp rates are shown to precisely quantify the plate motion of each segment in the ITRF reference frame. They are thus added back to the velocity maps before along‐track merging. Then, a simple referencing procedure implying only a constant and a tilt in azimuth adjusted for each track is constrained by horizontal GNSS data. It allows straightforward vertical and horizontal decomposition. The quality of the merged and referenced data sets can then be quantitatively assessed at 1 mm/yr by the differences between InSAR and GNSS velocities and by the InSAR residuals on track overlaps. The vertical velocity field is dominated at first order by the effect of permafrost degradation, while the horizontal velocity highlights strain accumulation across major strike‐slip faults in eastern Tibet.