Abstract: Autocatalysis from the by-product metal sulfides plays a critical role in the residual oil hydrotreating (RHT) process. However, it has not been considered to build the catalyst deactivation models, which probably is one important reason that the widely used S-type deactivation models are inaccurate in predicting some RHT processes' deactivation profiles. A three-stage catalyst deactivation model was first developed to fill this gap based on the mechanism inferred from the experimental and literature data. This model accounts for active site formation from by-product metal sulfides, deactivation due to active site coverage by coke formation and metal deposition, active site poisoning by highly-adsorbed species, active phase sintering, and diffusion resistance from the pore blockage at the same time, resulting in a function of dimensionless metals-on-catalyst. Then, the effectiveness of the proposed model was evaluated using the industrial data of an RHT unit and the experimental data from the literature, either in combination with reaction kinetics or independently. Results showed that RHT processes with clear autocatalytic effects may display different types of deactivation profiles from the traditional "S" shape. However, the proposed model was able to accurately track the entire deactivation curve of the RHT process and well predict the product properties. This approach yields valuable insights into the intricate autocatalytic effect that remarkably contributes to the performance modification of RHT catalysts. It is highly recommended that further research should be conducted on this topic, as it shows great potential to significantly advance catalyst and process development.