There is an established body of evidence indicating that a pupil's relative age within their school year cohort is associated with academic attainment throughout compulsory education. In England, autumn-born pupils consistently attain at higher levels than summer-born pupils. Analysis here investigates a possible channel of this relative age effect: ability grouping in early primary school. Relatively younger children tend more often to be placed in the lowest in-class ability groups, and relatively older children in the highest group. In addition, teacher perceptions of pupils' ability and attainment are associated with the child's birth month: older children are more likely to be judged above average by their teachers. Using 2008 data for 5481 English seven-year-old pupils and their teachers from the Millennium Cohort Study, this research uses linear regression modelling to explore whether birth month gradation in teacher perceptions of pupils is more pronounced when pupils are in-class ability grouped than when they are not. It finds an amplification of the already disproportionate tendency of teachers to judge autumn-born children as more able when grouping takes place. The autumn-summer difference in teacher judgements is significantly more pronounced among in-class ability grouped pupils than among non-grouped pupils. Given evidence that teacher perceptions and expectations can influence children's trajectories, this supports the hypothesis that in-class ability grouping in early primary school may be instrumental in creating the relative age effect.