Background Preterm birth is closely associated with neurocognitive impairment in childhood including increased risk for social difficulties. Eye tracking objectively assesses eye‐gaze behaviour in response to visual stimuli, which permits inference about underlying cognitive processes. We tested the hypothesis that social orienting in infancy is altered by preterm birth. Methods Fifty preterm infants with mean (range) gestational age (GA) at birth of 29+1 (23+2–33+0) weeks and 50 term infants with mean (range) GA at birth 40+2 (37+0–42+3) weeks underwent eye tracking at median age of 7 months. Infants were presented with three categories of social stimuli of increasing complexity. Time to first fixate (TFF) and looking time (LT) on areas of interest (AoIs) were recorded using remote eye tracking. Results Preterm infants consistently fixated for a shorter time on social content than term infants across all three tasks: face‐scanning (fixation to eyes minus mouth 0.61s vs. 1.47s, p = .013); face pop‐out task (fixation to face 0.8s vs. 1.34s, p = .023); and social preferential looking (1.16s vs. 1.5s p = .02). Time given to AoIs containing social content as a proportion of LT at the whole stimulus was lower in preterm infants across all three tasks. These results were not explained by differences in overall looking time between the groups. Conclusions Eye tracking provides early evidence of atypical cognition after preterm birth, and may be a useful tool for stratifying infants at risk of impairment for early interventions designed to improve outcome.