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.