Two processing centers providing radio occultation (RO) data from the same RO instrument CHAMP have been analyzed in order to estimate the magnitude of structural uncertainty and to test whether different RO data streams or different RO instruments can be combined easily to generate a long term climate monitoring data set. 10,000 CHAMP occultations processed at the centers were selected. The agreement in mean tangent point positions is about 15 km; it is mainly biased at low and mid-latitudes. Agreement in bending angle, refractivity, and dry temperature profiles is generally not sufficient for climate monitoring and requires further coordination among processing centers. Temperature biases of almost 5 K were found at 35 km altitudes, at lower altitudes biases are smaller but can still be around 1 K. The best bias agreement is found for the product closest to the raw data, bending angle.