Twenty-eight preterm infants of 28 weeks gestational age were observed four times over eight weeks in order to determine changes in their pain response. Both routine and sham heelstick procedures were used. Physiological (heart rate and oxygen saturations) and behavioural parameters (three upper facial actions) were used as outcomes. The responses to real heelstick were significantly greater than to sham heelstick for heart rate and all facial actions except one at 28 weeks gestational age, but not for oxygen saturation. The magnitude of response to both real and sham heelstick increased over time. Thus, the older the infant, the more robust and recognisable the response. Since even the youngest infants showed a differential response to pain, professionals caring for such infants need to be able to recognize their more subtle pain responses.