The work aims to improve the control performance of pulp concentration to overcome the large lag, non-linearity and time variability of pulp concentration control system. For the pulp concentration control problem, a modified internal model control was put forward. This method firstly introduced direct curve to change the original target control to process control, thus solving the contradiction between system speediness and overshoot. Then, the reference model of controlled object and given lag were introduced to take the deviation between guide output and reference model as control circuit A and the deviation between given lag and controlled object as control circuit B. Finally, based on the principle of internal model control, the internal model controllers were respectively designed for circuits A and B, and the sum of output from the two controllers was taken as the general controller. The pulp concentration system simulation platform with improved internal model control was built by Matlab simulation software and compared with the traditional control. The simulation results showed that the response speed, tracking performance, robustness and anti-interference characteristics of the system were better than the traditional control method. The effectiveness and feasibility of improved internal model control in pulp concentration control system are verified by simulation, and the comprehensive performance index of the system can be improved obviously.