Hodgkin (SPIE 6207(2006)) extended NVThermIP to be applicable to cold weather conditions. We also (IRPhys&Technol.51 (2008)520) later published an analysis of the effect of varying ambient temperature (T amb ) by modifying the inputs to NVTherm2002, and by using spectrally-weighted atmospheric transmission calculated from MODTRAN at different ambient temperatures and relative humidities (RH). We took into account the effects on the integration time and NETD, and we now account for the variation of ΔT with varying T amb , as Hodgkin has done. The overall trends are similar, but we have NVTherm, not NVThermIP. We vary the parameters associated with Johnson's criteria to obtain similar results. Note that diurnal, seasonal, climatic and microclimatic variations of relative humidity (RH) significantly impact the performance of thermal imagers, especially LWIR ones. We compare the performance of thermal imagers a horizontal mean-sea-level path in clear weather conditions for terrestrial imagers and ground targets/scenes in both LWIR and MWIR bands, as a function of the ambient temperature from -40°C to +40°C and also as a function of RH (30%, 50% and 70%). To understand the differences in the results reported by Hodgkin and our paper, we do a sensitivity analysis as a function of system and environmental parameters (f/#, RH, detection probability, spectral width etc). For one set of parameters, we observe that the range curves R LW and R MW intersect at more than one value of T amb and suggest an analogy to a 're-entrant phase'. We also analyze how motion blur affects the two bands, at different T amb .