Rubber fatigue crack growth tests are one method of characterising fatigue behaviour. Due to the possibility of using one specimen to measure fatigue at multiple load levels, it is often considered a faster and less expensive method compared to the time-consuming fatigue crack initiation tests. In this work, the possibility of a new specimen grip geometry for crack growth tests was investigated, enabling the use of specimens cut from compression moulded plates. Several variations of a crack growth predictor parameter for tests with different mean strain levels were investigated. In addition, the optimum variation of the crack growth angle with respect to the loading direction was evaluated using test measurements. The results show that the use of the simplified crack growth predictor can provide the same results as the commonly used tearing energy. Using the Euclidean length of the crack growth increments gives better results than their horizontal projection, with the possibility of using angle deviations up to 90° without major drawbacks. •A novel specimen grip geometry for planar tension specimens is presented.•Different variations of the tearing energy parameter have been investigated.•A modified tearing energy parameter has advantageous computational simplifications.•The effect of the crack growth angle on the scatter of the data was investigated.