The Transiting Exoplanet Survey Satellite (TESS) is NASAs latest space telescope dedicated to the discovery of transiting exoplanets around nearby stars. Besides the main goal of the mission, asteroseismology is an important secondary goal and very relevant for the high-quality time series that TESS will make during its two year all-sky survey. Using TESS for asteroseismology introduces strong timing requirements, especially for coherent oscillators. Although the internal clock on board TESS is precise in its own time, it might have a constant drift and will thus need calibration, or offsets might inadvertently be introduced. Here we present simultaneously ground- and space-based observations of primary eclipses of several binary systems in the Southern ecliptic hemisphere, used to verify the reliability of the TESS timestamps. From twelve contemporaneous TESS/ground observations we determined a time offset equal to 5.8 +/- 2.5 sec, in the sense that the Barycentric time measured by TESS is ahead of real time. The offset is consistent with zero at 2.3-sigma level. In addition, we used 405 individually measured mid-eclipse times of 26 eclipsing binary stars observed solely by TESS to test the existence of a potential drift with a monotonic growth (or decay) affecting the observations of all stars. We find a drift corresponding to sigma_drift = 0.009 +/- 0.015 sec/day. We find that the measured offset is of a size that will not become an issue for comparing ground-based and space data for coherent oscillations for most of the targets observed with TESS.