QSPR modeling was performed on 38 organic peroxides against onset temperatureThe mol. structures were optimized and frequencies were computed with DFT/b3lyp and 6-311g(d,p) basis set.Avogadro 1.2 was used to Convert gaussian output files to MDL SDFile format. 5666 mol. descriptors were calculated by Alvadesc software based on MDL SDfiles.The data set was randomly divided into training and test sets, and the number of samples contained in the training and test sets are 30 and 8, resp.A seven-parameter relationship was established by the stepwise multiple linear regression method.The correlation coefficient reaches 0.9843 and the RMSE for the training and test sets were 5.47 and 5.57, resp.Student t-value, VIF value and Durbin-Watson value indicated that all the independent variables had significant impact on dependent variable, and there was no multicollinearity between independent variables and no correlation between the residuals, resp.LOOCV, LMOCV, external validation, Application domain anal. showed that the resulting QSPR model was robust and reliable and could be used to predict the thermal decomposition temperature of organic peroxides.