Increased curriculum time allocated to reading might not be effective in raising achievement. Teachers need to closely monitor and manage both the quality and quantity of individualised reading of all their pupils for optimal effectiveness. 'Learning information systems' (LIS) for reading such as the 'Accelerated Reader' (AR) enable this through individualised computerised assessment of pupil comprehension of 'real books', with feedback to both pupil and teacher. This study explored the impact of AR on reading achievement in 13 schools of different types spread across the UK, the majority socio-economically disadvantaged. Participating pupils were aged 7-14 years. Pre-post norm-referenced gains in reading achievement were measured by group paper-reading tests and a computer-based adaptive reading test. The implementation integrity of AR was assessed by direct observation by researchers and through data generated by the programme itself. On both paper and computer-based reading tests, on aggregate pupils in the 13 schools gained in reading at abnormally high and statistically significant rates. Boys tended to show larger gains than girls on the paper test. However, implementation integrity was very variable. In particular, some teachers failed to intervene in response to AR data indicating that pupils were reading ineffectively. AR appears to have potential for raising reading achievement, but only if implemented appropriately.