Since 2004, atmospheric carbon dioxide (CO2) is being measured at the High Altitude Research Station Jungfraujoch by the division of Climate and Environmental Physics at the University of Bern (KUP) using a nondispersive infrared gas analyzer (NDIR) in combination with a paramagnetic O2 analyzer. In January 2010, CO2 measurements based on cavity ring-down spectroscopy (CRDS) as part of the Swiss National Air Pollution Monitoring Network were added by the Swiss Federal Laboratories for Materials Science and Technology (Empa). To ensure a smooth transition – a prerequisite when merging two data sets, e.g., for trend determinations – the two measurement systems run in parallel for several years. Such a long-term intercomparison also allows the identification of potential offsets between the two data sets and the collection of information about the compatibility of the two systems on different time scales. A good agreement of the seasonality, short-term variations and, to a lesser extent mainly due to the short common period, trend calculations is observed. However, the comparison reveals some issues related to the stability of the calibration gases of the KUP system and their assigned CO2 mole fraction. It is possible to adapt an improved calibration strategy based on standard gas determinations, which leads to better agreement between the two data sets. By excluding periods with technical problems and bad calibration gas cylinders, the average hourly difference (CRDS – NDIR) of the two systems is -0.03 ppm ± 0.25 ppm. Although the difference of the two data sets is in line with the compatibility goal of ±0.1 ppm of the World Meteorological Organization (WMO), the standard deviation is still too high. A significant part of this uncertainty originates from the necessity to switch the KUP system frequently (every 12 min) for 6 min from ambient air to a working gas in order to correct short-term variations of the O2 measurement system. Allowing additional time for signal stabilization after switching the sample, an effective data coverage of only one-sixth for the KUP system is achieved while the Empa system has a nearly complete data coverage. Additionally, different internal volumes and flow rates may affect observed differences.