An analysis of 206,422 atmospheric refractivity profiles observed by the CHAMP georesearch satellite between 14 May 2001 and 30 June 2005 reveals significant biases compared to European Centre for Medium‐Range Weather Forecasts meteorological fields at altitudes below 5 km. The mean bias decreases down to −2% at altitudes below 2 km; in the Amazon region, positive biases exceeding +1% are observed. In order to identify bias contributions caused by the receiver signal tracking process an end‐to‐end simulation study implementing different signal tracking modes was performed. The end‐to‐end simulations are based on 1992 radiosonde profiles obtained regularly aboard R/V Polarstern since December 1982 and were conducted with four receiver models using closed‐loop, fly wheeling, and open‐loop signal tracking methods. The simulation results confirm that open‐loop tracking yields significantly smaller biases and standard deviations of the fractional refractivity errors compared to fly wheeling enabled receivers. In addition, we analyze closed‐loop tracking with a second‐order loop and demonstrate that similar reductions in biases and standard deviations can be obtained.