Current prediction methods for growth of disbonds under fatigue loading are generally based on a correlation with either the maximum strain energy release rate (SERR) or the SERR range. This paper highlights some issues with this approach. In particular, it is argued that the maximum SERR or the SERR range alone do not give sufficient information to uniquely characterize the driving force for crack growth. Furthermore it is argued that the relationship between crack growth rate and loss of strain energy should be considered on the scale of the entire load cycle. By means of disbond growth experiments it is shown that there is indeed a very strong correlation between the crack growth rate and the strain energy lost during a fatigue cycle. Unlike methods based on the SERR, this correlation is not affected by the R-ratio. Based on the found correlation a possible basis for a new approach to disbond growth prediction is suggested.