The fatigue crack growth(FCG) experiment and modeling method for rubbers used in isolators under variable amplitude loadings are carried out using an edge-crack pure shear specimen. The commonly used data processing techniques for getting the crack growth rate(crank length versus number of cycles) in metal materials, the secant method or the local incremental polynomial method, are not applicable for computing the crack growth rate of rubber materials. A power function is proposed to fit the measured crack growth length and number of cycles using the least-squares technique to avoid the deficiencies of the conventional data processing methods. The crack growth rate is then calculated from the determined power function, so as to establish a FCG prediction model for the rubber used in isolators. A dumbbell specimen made of the same rubber compound as the pure shear specimen is then manufactured and is used to carry out the tensional fatigue experiment. The comparisons between the measured tensional fatigue life of the dumbbell specimen and that evaluated from the established FCG model validates the proposed data processing method for FCG data of rubbers used in isolators under variable amplitude loads.