There have been many previous studies on the tearing of rubber materials. The behaviour is typically characterised using a fracture mechanics approach whereby the rubber has a geometrically independent relationship between crack growth rate during tearing versus strain energy release rate. This approach works well under conditions of steady tearing as the crack growth rate is easy to measure. However, this approach is much harder to interpret under conditions where the rubber exhibits discontinuous crack growth behaviour such as knotty tearing or stick slip tearing. Unfortunately, these are the most common tear conditions observed in practice for filled rubbers as well as for some unfilled rubbers, especially those such as natural rubber that are capable of strain induced crystallisation. Under these conditions it is not clear what the actual crack growth rate is as the value typically given results from the average of a very rapid tear rate and a zero velocity tear rate. The approach developed here for the first time is to characterise the rate of increase in the strain energy at the crack tip just immediately before the onset of tearing in a trouser tear test piece. This is measured directly as the time derivative of the strain energy release rate T • . In this work the relation between this and the critical strain energy release rate T crit to propagate the tear is examined. Having developed a new technique to characterise the tear behaviour, it is then evaluated using a range of different materials, including both natural rubber (NR) and styrene butadiene rubber (SBR). This allows the effect of the strain induced crystallisation observed in NR to be evaluated. The level of cross linking is varied to modify the kinetics of the strain induced crystal formation, and carbon black is incorporated into some of the NR and SBR compounds. The new technique is seen to provide a useful way of examining the different reinforcing mechanisms exhibited by all these different rubber materials.