The neuropsychological test scores of 47 children with traumatic head injury (THI), aged 9-14 years, were evaluated to determine the relative contributions of severity of injury (length of coma) and age on performance. Stepwise regression analyses revealed that age consistently explained a statistically significant proportion of the variance on the Trail Making Test (TMT) and Finger Tapping Test (FTT), even after accounting for injury severity. However, logistic regression analyses indicated that currently available age-based norms for these tests did not improve classification of mild-moderate vs. severe injuries, as compared to a system based exclusively on raw scores. It is concluded that, whereas age is an important consideration when assessing children with THI, available age-based norms for some common paediatric neuropsychological tests do not appear to be sufficiently stratified or robust to fully capture these age effects. It is suggested that more comprehensive age-referenced norms, based on larger and more adequately stratified samples, are needed. In addition, some practical implications of these findings are discussed.