The dependence of gas temperature on gas flow through a direct current, slot-type, atmospheric pressure microplasma in helium or argon was investigated by a combination of experiments and modelling. Spatially resolved gas temperature profiles across the gap between the two electrodes were obtained from rotational analysis of N 2 (C 3 u → B 3 g) emission spectra, with small amounts of N 2 added as actinometer gas. In Ar/N 2 discharges, the N 2 (C 3 u v = 0 → B 3 g v = 0) emission spectra were fitted with a two-temperature population distribution of the N 2 (C) state, and the gas temperature was obtained from the 'low temperature' component of the distribution. Under the same input power of 20 kW cm −3 , the peak gas temperature in helium (∼650 K) was significantly lower than that in argon (over 1200 K). This reflects the much higher thermal conductivity of helium gas. The gas temperature decreased with increasing gas flow rate, more so in argon compared with helium. This was consistent with the fact that conductive heat losses dominate in helium microplasmas, while convective heat losses play a major role in argon microplasmas. A plasma-gas flow simulation of the microdischarge, including a chemistry set, a compressible Navier-Stokes (and mass continuity) equation and a convective heat transport equation, was also performed. Experimental measurements were in good agreement with simulation predictions.