In the framework of the EMPIR project traceRadon, stable atmospheres with low-level radon activity concentrations have to be produced for calibrating radon detectors designed to measure outdoor air activity concentrations. The traceable calibration of these detectors at very low activity concentrations is of special interest to the radiation protection, climate observation, and atmospheric research communities.Radiation protection networks (such as the EUropean Radiological Data Exchange Platform (EURDEP)) and atmospheric monitoring networks (such as the Integrated Carbon Observation System (ICOS)) need reliable and accurate radon activity concentration measurements for a variety of reasons, including: the identification of Radon Priority Areas (RPA); improving the sensitivity and reliability of radiological emergency early warning systems (Melintescu et al., 2018); for more reliable application of the Radon Tracer Method (RTM) to estimate greenhouse gas (GHG) emissions; for improved global “baseline” monitoring of changing GHG concentrations and quantification of regional pollution transport (Chambers et al., 2016), (Chambers et al., 2018); and for evaluating mixing and transport parameterisations in regional or global chemical transport models (CTMs) (Zhang et al., 2021), (Chambers et al., 2019).To achieve this goal, low activity sources of radium with a variety of characteristics were produced using different methods. Sources ranging from MBq 226Ra down to several Bq 226Ra were developed and characterised during the evolution of production methods, and uncertainties below 2 % (k=1) were achieved through dedicated detection techniques, even for the lowest activity sources. The uncertainty of the lowest activity sources was improved using a new online measurement technique for which the source and detector were combined in the same device. This Integrated Radon Source Detector device, henceforth an IRSD, reaches a counting efficiency approaching 50 % through detection under quasi 2π sr solid-angle. At the time of this study the IRSD was already produced with 226Ra activities between 2 Bq and 440 Bq.To compare the working performance of the developed sources (i.e., to establish a reference atmosphere), study the stability of the sources, and to establish traceability to national standards, an intercomparison exercise was carried out at the PTB facility. Here we present the various source production techniques, the determination of their radium activity, and determination of their radon emanation (including assigned uncertainties). This includes details of the implementation of the intercomparison set-up, and a discussion of the results of the source characterisations.