Abstract Cleavage fracture toughness data often exhibits a considerable amount of scatter and strong dependencies on temperature, loading rate, pre-straining, etc. due to the highly localized nature of the failure mechanism and microstructural inhomogeneity in the ductile-to-brittle transition (DBT) region of a material. Beremin's model is popularly used by the researchers to predict the cleavage fracture probabilities. However, the temperature dependency of the Weibull parameters is an issue which is still being debated worldwide. Usually, elasto-plastic analysis is carried out to determine the Weibull stress and hence the effect of ductile crack growth prior to cleavage fracture is not considered. In this work, it is demonstrated that the temperature dependency of the parameters is observed due to inability of the models to predict the above pre-cleavage ductile crack growth. The authors have developed a combined nonlocal (mesh independent) ductile damage and cleavage fracture model to predict the fracture toughness transition curve of ferritic steels. Extensive experiments have been conducted on compact tension specimens in the DBT region and the predicted results have been compared with those of experiment to demonstrate the merits of the new model.