Granite residual soils are characterized by poor water stability, high permeability coefficients, and susceptibility to damage by rainfall scour when used for roadbed filling. To explore the effect of microbial induced calcite precipitation （MICP） technique on the permeability coefficient of granite residual soils and the mechanism of action, the permeability coefficient of remodeled granite residual soil specimens before and after MICP curing was measured, and the calcium carbonate content, nuclear magnetic resonance scan, scanning electron microscope scan and X-ray diffraction analysis were performed on the specimens before and after curing. The results showed that 1) The MICP technique can reduce the permeability coefficient of granite residual soil specimens by two orders of magnitude, in the test range, the order of the influence degree of each factor on the permeability coefficient from large to small is: cement concentration, number of cycles, infusion volume, and the best effect is achieved when the infusion volume is 40 mL, the concentration of cement is 1.0 mol/L and the curing cycle is 16 times. 2) The precipitation generated by curing fills in the between the pores of the granite residual soil, resulting in poor connectivity between the pores and a decrease in porosity, and overall the higher the calcium carbonate content near the injection end, the better the curing effect. 3) The clustered calcite calcium carbonate effectively cemented the originally loose soil particles, blocked the seepage channels between the soil particles pores, and reduced the permeability coefficient of the specimen.