The Glasgow Coma Scale is widely accepted as a measure of impaired consciousness both in clinical practice and in research. In its standard form, the scale is inapplicable to infants and children below the age of 5 years. We have devised a paediatric coma scale, which recognises that the expected normal verbal and motor responses must be related to the patient's age. This scale has been routinely used in a paediatric teaching hospital over a 10-year period and has appeared to be satisfactory, although it is less sensitive to changes in the conscious level than the adult scale. We have assessed the consistency of observer evaluation, using video tapes of infants and children exhibiting varying levels of consciousness: when conducted after formal instruction, assessments were reasonably consistent.