The invention discloses an integrated circuit X-value sensitivity prediction method based on a random forest model. The method comprises the following steps: 1, comprehensively generating a circuit gate-level netlist for an original circuit netlist; 2, calculating the level and depth of each logic gate through a circuit gate level netlist; 3, carrying out region division on the circuit according to different topological structures when each input port is an X value; 4, extracting the structural features of the circuit when different input ports are X values according to the importance of different regions; 5, calculating test coverage rate loss when different input ports are X values through an ATPG tool, and forming a sample data set in combination with the extracted structural features;and 6, training and predicting the sample data set through a random forest model. According to the method, the test coverage rate loss caused when different input ports in the circuit are X values canbe quickly and accurately predicted, so that the purposes of reducing the hardware overhead when the circuit design protects the X value module and improving the circuit test quality are achieved.