The uncertainty in determining the volatility behaviour of organic particles from thermograms using calibration curves and a kinetic model has been evaluated. In the analysis, factors such as re-condensation, departure from equilibrium and analysis methodology were considered as potential sources of uncertainty in deriving volatility distribution from thermograms obtained with currently used thermodenuder designs. <br><br> The previously found empirical relationship between <i>C</i>* (saturation concentration) and <i>T</i>₅₀ (temperature at which 50% of aerosol mass evaporates) was theoretically interpreted and tested to infer volatility distributions from experimental thermograms. The presented theoretical analysis shows that this empirical equation is in fact an equilibrium formulation, whose applicability is lessened as measurements deviate from equilibrium. While using a calibration curve between <i>C</i>* and <i>T</i>₅₀ to estimate volatility properties was found to hold at equilibrium, significant underestimation was obtained under kinetically-controlled evaporation conditions. Because thermograms obtained at ambient aerosol loading levels are most likely to show departure from equilibrium, the application of a kinetic evaporation model is more suitable for inferring volatility properties of atmospheric samples than the calibration curve approach; however, the kinetic model analysis implies significant uncertainty, due to its sensitivity to the assumption of "effective" net kinetic evaporation and condensation coefficients. The influence of re-condensation on thermograms from the thermodenuder designs under study was found to be highly dependent on the particular experimental condition, with a significant potential to affect volatility estimations for aerosol mass loadings >50 μg m⁻³ and with increasing effective kinetic coefficient for condensation and decreasing particle size. These results show that the geometry of current thermodenuder systems should be modified to prevent re-condensation.