This paper describes a randomised controlled trial conducted with 10 secondary schools in England to evaluate the impact and feasibility of Fresh Start as an intervention to help new entrants with low prior literacy. Fresh Start is a synthetic phonics programme for small groups of pupils, here implemented three times per week over 22 weeks. The intervention was led by the schools themselves and assessed in cooperation with the independent evaluators. A group of 433 year 7 pupils (first year of secondary school) were identified by schools as having literacy attainment below "secure" KS2 level 4, and individually randomised to a treatment group or a waiting-list control. The pupils were assessed via GL's New Group Reading Test. Missing data at pre- and post-test amounted to 3% of the total. The overall "effect" size in terms of gain scores from pre- to post-test was +0.24, and this was repeated in a sub-group analysis involving only FSM-eligible pupils. However, there was some imbalance between the two groups at the outset, and this must lead to a slight caution about the findings, and to some doubts about whether one or more schools unwittingly subverted the randomisation. Other than this, the aggregated trial shows that schools can conduct evaluations of their own interventions with firm guidance from experts, and under favourable conditions such as individual randomisation and lack of extended involvement by developers with a conflict of interests.