A quantitative structure–property relationship (QSPR) study is performed to develop mathematical models for the prediction of the flash point (FP) of organosilicon compounds from their molecular structures. Various kinds of molecular descriptors were calculated to represent the molecular structures of organosilicon compounds, such as topological, charge, and geometric descriptors. The genetic algorithm combined with multiple linear regression (GA-MLR) is employed to a select optimal subset of descriptors that have a significant contribution to the overall FP property. The model with the best result is a five-variable multilinear model, which showed high prediction ability when the obtained root mean square error and average absolute error for the external test set were 14.11 and 11.1 K, respectively. The model was further compared with other previously published methods. The results indicate the superiority of the presented model and reveal that it can be effectively used to predict the FP of organosilicon compounds with only the knowledge of molecular structures. This study can provide a new way for predicting the FP of organosilicon compounds for the engineering field. Copyright © 2012 John Wiley & Sons, Ltd.