Purpose: We developed and evaluated in a randomized controlled trial a computerized speechreading training program to determine (a) whether it is possible to train speechreading in deaf children and (b) whether speechreading training results in improvements in phonological and reading skills. Previous studies indicate a relationship between speechreading and reading skill and further suggest this relationship may be mediated by improved phonological representations. This is important since many deaf children find learning to read to be very challenging. Method: Sixty-six deaf 5- to 7-year-olds were randomized into speechreading and maths training arms. Each training program was composed of a 10-min sessions a day, 4 days a week for 12 weeks. Children were assessed on a battery of language and literacy measures before training, immediately after training, and 3 months and 11 months after training. Results: We found no significant benefits for participants who completed the speechreading training, compared to those who completed the maths training, on the speechreading primary outcome measure. However, significantly greater gains were observed in the speechreading training group on one of the secondary measures of speechreading. There was also some evidence of beneficial effects of the speechreading training on phonological representations; however, these effects were weaker. No benefits were seen to word reading. Conclusions: Speechreading skill is trainable in deaf children. However, to support early reading, training may need to be longer or embedded in a broader literacy program. Nevertheless, a training tool that can improve speechreading is likely to be of great interest to professionals working with deaf children.