For efficient inherently safe design, awareness of all available options at the appropriate decision-making moment is key. Simulations offer both timely availability and large screening possibilities. Therefore computer-aided product design gained in popularity during the recent years and models of various physico-chemical properties were developed. Yet, predictions of safety related data are still limited. Hence, thermal stability derived from Differential Scanning Calorimetry (DSC) is analysed and simulated with two molecular-based modelling approaches: Group Contribution Method (GCM) and Quantitative Structure-Property Relationships (QSPR). Predictive models are developed and evaluated over fitting and predictive abilities for five properties extracted from the DSC curves.