Background:  We evaluate the effectiveness of three theoretically motivated programmes for the teaching of reading to four‐year‐old children. Method:  Four hundred and ten children, of pre‐kindergarten age, in 20 UK Reception‐year classrooms were divided into four matched groups and randomly assigned to one of three experimental teaching conditions, Reading with Rhyme, Reading with Phoneme, Reading with Rhyme and Phoneme, or to a taught control condition (Reading). In each condition the Reading element contained a strong phonic component. Results:  There were no selective effects of the different experimental teaching programmes for normally developing children. However, for those children identified as being at risk of reading failure, training in phoneme skills resulted in selective gains in phoneme awareness and in reading skills. Conclusions:  A reading programme that contains a highly structured phonic component is sufficient for most 4.5‐year‐old children to master the alphabetic principle and to learn to read effectively, without additional explicit phonological training. In contrast, for young children at risk of reading delay, additional training in phoneme awareness and linking phonemes with letters is beneficial.