The effects of variations in word length on serial recall were examined in a group with Down's syndrome, a group with severe learning difficulties of mixed aetiology, and normal children. A split-plot analysis of variance was performed on the mean number of words recalled in correct serial position on each list, in which the factors were subject type, mental age and word length. The main effect of subject type was further explored using a Tukey HSD test. This revealed a significant difference in recall between the normal and Down's syndrome groups and also between the normal and mixed aetiology-severe learning difficulties group. The results of this experiment reveal very clear differences between the effects of word duration on serial recall in normal children and children with severe learning difficulties of different aetiologies and at a range of different mental ages.