A new scale for assessment of overall responsiveness, the Reaction Level Scale (RLS 85), which has been shown to have better reliability than the Glasgow Coma Scale (GCS), has been tested in four neurosurgical departments regarding inter-observer agreement and coverage i.e. the proportion of patients that could be assessed by the scale. In a carefully designed study 51 observers pairwise performed 164 tests on 88 patients. Reliability was studied by the Kappa method, which is defined as inter-observer agreement corrected for agreement by chance. The inter-observer agreement measured as overall Kappa was good (K = 0.69 +/- 0.05) and there were no significant differences between the departments, professional categories or aetiologies. Regarding the separate RLS 85 levels the Kappa values were above 0.65, except for withdrawing (K = 0.51) and flexor responses (K = 0.55). There was good inter-observer agreement on coma (K = 0.71). In conclusion, the RLS 85 proved to be easily learnt, it showed full coverage without pseudoscoring, and it was used in a consistent way by doctors, nurses and assistant nurses of four different neurosurgical departments in two Scandinavian countries.