The Centre for Metrology and Accreditation (MIKES) is developing a temperature–humidity calibration system for radiosondes. The target minimum air temperature and dew-point temperature are -80∘C and -90∘C, respectively. When operating in this range, a major limiting factor is the time of stabilization which is mainly affected by the design of the measurement chamber. To find an optimal geometry for the chamber, we developed a numerical simulation method taking into account heat and mass transfer in the chamber. This paper describes the method and its experimental validation using two stainless steel chambers with different geometries. The numerical simulation was carried out using Comsol Multiphysics simulation software. Equilibrium states of dry air flow at -70∘C with different inlet air flow rates were used to determine the geometry of the chamber. It was revealed that the flow is very unstable despite having relatively small Reynolds number values. Humidity saturation abilities of the new chamber were studied by simulating water vapor diffusion in the chamber in time-dependent mode. The differences in time of humidity stabilization after a step change were determined for both the new chamber model and the MIKES Relative Humidity Generator III (MRHG) model. These simulations were used as a validation of the simulation method along with experimental measurements using a spectroscopic hygrometer. Humidity saturation stabilization simulations proved the new chamber to be the faster of the two, which was confirmed by experimental measurements.