The European Space Agency is sending its first rover to Mars as part of the ExoMars programme. There is large international cooperation with Roscosmos, with some contributions from NASA and there are two launches as part of the programme. The 2016 mission, which consists of a Trace Gas Orbiter (TGO) and an EDL Demonstrator Module (EDM), launched in March 2016 and the Rover and Surface Platform Mission to be launched in 2020 which consists of a Carrier Module (CM) and a Descent Module (DM) with a rover and a stationary landing platform. The cold Martian nights pose a significant challenge for the rover’s thermal control - keeping the rover service module warm while minimising the electrical power consumption. The rover service module is thermally decoupled from the environment radiatively, utilising low emissivity finishes, conductively, using low conductance supports and harness connectors and convectively with a carbon dioxide gas-gap. Methods of modelling radiation and conduction are well understood, but modelling of convection in low pressure and low Rayleigh number environments, however, is not. To better understand the underlying mechanisms behind the heat flow through the gas, testing has been carried out in three configurations with gaps and temperature gradients representative of the rover design. Testing was performed in vacuum, in air at Earth ambient pressure and also in CO2 and N2 at pressures representative of Martian Rayleigh numbers. These configurations were then modelled, with five different approaches taken to modelling the gas couplings; pure conduction with multiple gas nodes, convection into free space with one gas node per gap, convection into free space with multiple gas nodes, limited parallel plates without gas nodes and temperature dependant limited parallel plates without gas nodes. The test results are compared to each modelling method and the use of each is evaluated.