A simple and accurate group contribution method to predict the flash point of pure organic compounds is presented. The model was developed using a large data set of 1660 components from different material classes. The prediction of the flash point requires only the molecular structure and normal boiling point as input information. To achieve better accuracy an analysis of the relation between the flash point temperature and the normal boiling point has been carried out. The predictive capability and reliability of the new model have been demonstrated on a test set of 332 compounds with an average absolute relative deviation of 1.62% and an average absolute deviation of 5.7 K. The performance of the developed model has been compared with several other currently used group-contribution models and the results show that the new model is more reliable than existing models. (C) 2017 Elsevier B.V. All rights reserved.