We present a comprehensive new pyrolysis data set for a maturity series (0.55–2.86% vitrinite reflectance) of Pennsylvanian coals (Westphalian, Ruhr Basin). The broad similarity of the organofacies was established by elemental and petrographic data. Comparison of maximum temperatures reached by the coals was based on commonly applied algorithms (Easy%Ro, Easy%RoDL, Basin%Ro). These temperatures were finally compared to predicted temperatures of petroleum generation from the same coals in order to test the reliability of kinetic data for numerical petroleum generation reconstruction from coal. Kinetic parameters were determined based on data from open-system pyrolysis performed on two instruments (Rock-Eval™ and Source Rock Analyzer™) and basin modelling was performed using PetroMod™ software and a simple, rapid burial and heating history (about 20 °C/million years), which is similar to the rapid heating which caused natural maturation of these coals.With increasing maturity, pyrolysis curves shift towards higher temperatures, no matter which heating rate is used. However, pyrolysis curves of higher maturity samples are not embedded within the more extended and larger pyrolysis curves of low maturity coals. HI values and H/C ratios do not significantly decrease over the maturity range from 0.55 to 0.8–0.9% vitrinite reflectance. These observations indicate significant restructuring of organic matter in coals and only limited petroleum generation and expulsion within this maturity range.The kinetic parameters were derived from single (5 °C/min) and multiple (0.7, 2, 5, 15 °C/min) heating rate pyrolysis tests. After temperature correction, kinetic parameters obtained by the two instruments differ only slightly.Kinetics based on multiple heating rates using a variable frequency factor and variable activation energies provide the best mathematical fit to the laboratory data but geologically inconsistent predictions for natural petroleum generation, i.e. no increase in predicted generation temperatures with increasing sample maturity. In contrast, kinetics based on a constant, physically meaningful frequency factor and a calculated range of activation energies provide more consistent results for the maturity sequence. Interestingly, single-ramp kinetics (5 °C/min) show similar results, i.e. increasing calculated temperatures for petroleum generation with increasing maturity.For the lowest maturity sample, petroleum generation temperatures (10–75% conversion) are higher than calculated maximum rock temperatures, while this is not the case for all higher maturity samples, for which much of the predicted petroleum generation occurs at “too low” temperatures, i.e. temperatures clearly lower than calculated maximum rock temperatures. Maximum temperatures predicted by the Easy%RoDL approach are about 10 °C closer to the lower limit of petroleum generation than those calculated by Basin%Ro. Thus, kinetic data can act as a useful tool for calculating petroleum generation in petroleum system modelling providing clues towards the thermal stability of kerogen. However, such data are far from describing petroleum generation in nature exactly and quantitatively, especially for coals.