The effect of complexity level of a lumped kinetic model for heavy residue hydroconversion on estimated values of kinetic parameters was investigated in this work by imposing constraints for the parameter estimation algorithm of a complex six-lump kinetic model and deriving a simpler modified model from the complex model. Kinetic analysis was performed using available experimental data reported in the literature from a study on hydrocracking of Chinese Gudao vacuum residue in a bench-scale reactor using ammonium phosphomolybdate (APM) as a dispersed catalyst. The kinetic models also included coke formation reactions that had previously been ignored by most investigators due to the rather small amounts of coked formed under moderate reaction temperatures. Three constrained levels were incorporated into the parameters estimation algorithm of the kinetic model and the Average Absolute Errors (AAE) between experimental and predicted product mass fractions for each model were compared. It was found that while AAE increased by imposing additional constraints, more logical kinetic parameters were obtained. The AAE% of the last constrained level of the complex model was 9.2%. A simpler modified model was then defined based on the results of the original complex model. The AAE% after sensitivity analysis of the estimated Arrhenius parameters for the modified model without imposing any restrictions on parameter values was 8.8%. (C) 2014 Elsevier Ltd. All rights reserved.