The need for scales describing long-term consequences of injuries is growing. The introduction of the Injury Impairment Scale (IIS) is therefore an important step towards a more extended description of injuries, and an important complement to the AIS. This paper presents studies of the validity and reliability of the IIS for injuries that occured in Sweden in the late eighties and early nineties. The criteria for the IIS were compared to the settlements of an insurance company, on a single diagnosis level. It was found that the IIS scores did not reflect the outcomes from the injuries that had occured. This was true both for the level of impairment as well as the range of outcomes in real life. It was concluded that IIS at this stage cannot predict the number of impairments, as well as the probable levels of impairment. It was proposed that the impairment ratings should be based on both the probability of impairment as well as the level of impaiment.