Purpose. Enhanced Thinking Skills (ETS) has been the most widely delivered cognitive skills programme in the prisons of England and Wales. Four quasi‐experimental outcome studies have produced mixed results, a qualitative survey of offenders’ and facilitators’ experience on the programme proved useful in programme refinement, and a study using random allocation provided evidence that ETS impacts significantly and positively on short‐term attitudinal change. This study aims to make a further contribution, using another methodology, to the accumulation of evidence. Methods. This was a real‐world evaluation, comparing the reconviction outcomes of the population of 17,047 ETS participants in custody from 2000 to 2005 with a national cohort of 19,792 prisoners released over the same period. Results. Overall, prisoners who had attended ETS were found to reoffend at a rate 6.4 percentage points less than the cohort (rising to 7.5 percentage points for programme completers) and 9.5 percentage points less than the predicted rate. In all but the very highest risk group and in every sentence length band, the reoffending outcomes for ETS participants were significantly better than for prisoners in the cohort. Conclusions. It is argued that this non‐experimental methodology makes a contribution to the ‘What Works’ evidence.