Introduction Treatment trials in paediatric stroke will require valid and robust assessment of outcome. Although the Modified Rankin Scale (MRS) has been widely used in studies of adults, there are few validation studies, for example, of interobserver reliability, of this or other outcome measures following childhood stroke and few data on whether vascular occlusion, which might be the target of thrombolysis treatment predicts outcome.

Method Children admitted following ischaemic or haemorrhagic stroke to a regional UK centres were assessed at final follow-up, either directly or by questionnaire. Two paediatric neurological physiotherapists (PNP) independently derived scores on the MRS, range 0–5, as previously used after paediatric stroke, to calculate inter-observer reliability. Clinical and radiological predictors of MRS>3 were analysed using logistic regression.

Results 89 patients (44 boys) who had had an arterial ischaemic stroke at the age of 4.9 (0.1–17) years were assessed at the median age of 9.5 (2–24) years. Inter-observer agreement for MRS (PNP1 median 3 (1–5), PNP2 median 3 (0–5)) was good (κ 0.68; p<0.0005). Forty-seven (53%) had MRS>3, which was predicted by symptomatic aetiology (OR 3.6, 95% CI 1.3 to 9.8), independently of age and vascular occlusion which were predictors in univariable analysis.

Conclusion Although interobserver variability for the MRS is good, disparities between observers resulted from imprecise categories open to assessor interpretation about “age-appropriate activities” and “pre-existing disabilities”. Children with pre-existing disorders have poor outcome after stroke and might be candidates for trials of thrombolysis, although a larger data set is needed to determine whether vascular occlusion is an independent predictor of outcome.