Background Intervention studies for children at risk of dyslexia have typically been delivered preschool, and show short‐term effects on letter knowledge and phoneme awareness, with little transfer to literacy. Methods This randomised controlled trial evaluated the effectiveness of a reading and language intervention for 6‐year‐old children identified by research criteria as being at risk of dyslexia (n = 56), and their school‐identified peers (n = 89). An Experimental group received two 9‐week blocks of daily intervention delivered by trained teaching assistants; the Control group received 9 weeks of typical classroom instruction, followed by 9 weeks of intervention. Results Following mixed effects regression models and path analyses, small‐to‐moderate effects were shown on letter knowledge, phoneme awareness and taught vocabulary. However, these were fragile and short lived, and there was no reliable effect on the primary outcome of word‐level reading. Conclusions This new intervention was theoretically motivated and based on previous successful interventions, yet failed to show reliable effects on language and literacy measures following a rigorous evaluation. We suggest that the intervention may have been too short to yield improvements in oral language; and that literacy instruction in and beyond the classroom may have weakened training effects. We argue that reporting of null results makes an important contribution in terms of raising standards both of trial reporting and educational practice.