Change in facial expression over a fixed time after a noxious stimulus is the key measure used to calculate pain scores in preterm and newborn infants. We hypothesised that the latency of facial motor responses would be longer in the youngest premature infants and that behavioural scoring methods of pain may need to take this into account. One hundred and seventy-two clinically required heel lances were performed in 95 infants from 25 to 44 weeks postmenstrual age (PMA). Sixty-four percentage of the heel lances evoked a change in facial expression. Change in facial expression was observed in infants across the whole age range from 25 weeks PMA and the latency to the facial expression response ranged from 1 to 17s. Latency to facial expression change was dependent on the infants' PMA at the time of the heel lance. Infants below 32 weeks PMA had a significantly longer latency to change in facial expression than older infants (54% increase in infants below 32 weeks; p < 0.001). Sleep state and presence of brain damage (IVH grades 1-4) did not significantly increase the latency (p > 0.05 for each variable). Intravenous morphine at the time of the heel lance significantly increased the latency to facial expression response (p < 0.001) but the analysis shows that latency is highly dependent on PMA independent of morphine administration. These findings highlight developmental changes underlying infant behaviour that are critically important if pain scores are to be correctly interpreted.