A growing body of research highlights the importance of non-cognitive skills as determinants of young people's cognitive outcomes at school. However, little evidence exists about the effects of policies that specifically target students' non-cognitive skills as a way to improve educational achievements. In this paper, we shed light on this issue by studying a remedial education programme aimed at English secondary school pupils at risk of school exclusion and with worsening educational trajectories. The main peculiarity of this intervention is that it solely targets students' non-cognitive skills – such as self-confidence, locus of control, self-esteem and motivation – with the aim of improving pupils' records of attendance and end-of-compulsory-education (age 16) cognitive outcomes. We evaluate the effect of the policy on test scores in standardized national exams at age-16 using both least squares and propensity-score matching methods. Additionally, we exploit repeated observations on pupils’ test scores to control for unobservables that might affect students’ outcomes and selection into the programme. We find little evidence that the programme significantly helped treated youths to improve their age-16 test outcomes. We also find little evidence of heterogeneous policy effects along a variety of dimensions.