We propose an experimental and theoretical investigation of the effect of heat exchange through air upon the operation of on-chip test microstructures for heat conductivity measurements on thin films. By comparing measurements performed in vacuum (the usual method applied in most of the literature) and in air, a difference is devised on the determined thermal conductivity values in the two cases, depending on the thermal characteristics of the layer of interest. This discrepancy is interpreted by carefully evaluating the results of thermal simulations of the test structures' operation and explained in terms of a difference in thermal exchange through the gas phase induced by the presence of the measured layer on the test structure.