We present vertical temperature profiles derived by two different retrieval methods from nighttime radiation measurements performed by VIRTIS(M)-VEx (Visible and Infrared Thermal Imaging Spectrometer, M channel—Venus Express). The Bayesian approach to the optimal estimation method and the relaxation method are applied in this study. This is a first attempt to present and compare results obtained from two independent methods. It allows us to be more convinced of our interpretation.After comparison of temperature profiles we conclude that both retrieval methods are able to sound the atmospheric layers higher than 59 km (< 150 mb) with a single-standard deviation error of less than 5 K at the bottom and even deeper for some measurements down to 56 km but with error larger than 10 K. The best consistency between temperature profiles obtained from the two algorithms with the standard deviation less than 2.5 K is observed for atmospheric layers between 70 km–77 km (15–3.5 mb).In our conclusion we have no preference for any approach. Two methodologies are of equivalent value. Both methods resolve temperature inversions at high altitudes (∼84 km), the quality of fits for all observations is equally well. Only for the Bayesian approach, the retrieval uncertainty above 62 km up to 95 km is less than 2 K. A disadvantage of this method is the time-consuming calculation of weighting functions.The atmospheric temperatures over the “cold collar” region located at 60°S–75°S (60–70 km) are ∼10 K smaller than for latitudes poleward of 75°S (polar region). The cold collar region is seen very clearly in our results for both methods.