An accurate prediction of the fatigue life of a rubber product is based on a criterion. Since strain can be measured directly in rubber components during a test, it would help design engineers improve product performance effectively if the strain‐based damage criteria can achieve high accuracy. Hence, the experimental data on the AE42 and AE2 specimens were utilized to test the suitability of the proposed damage criterion for rubber materials. The criterion includes three principal components determined by both a strain range and a current strain. Data from AE42 specimens, made from styrene‐butadiene rubber, contained 40 cases, that is, tension, torsion, and tension‐torsion loadings. An S‐N curve was obtained, with a scatter band of 0.7 and a coefficient of determination R2 of 0.81. The curve could be useful at the design stage. A sample of AE2, made from natural rubber filled with carbon black, showed the location and orientation of the crack in a nonproportional case. The predicted location of the fatigue crack was consistent with the experimental observation and the crack orientation matched the experimental measurement, that is, the predicted angle of 33.4° versus the measurement of approximately 40°. It would be possible to combine the proposed approach with the critical plane method in nonproportional loadings to save significant CPU time: using the proposed approach to find the critical loading range and then using the critical plane method to find the maximum value of the required damage variables. More engineering cases are needed to further verify this proposed criterion.