Background In England, there is a growing need to improve the lives of secondary school students who are defined as disadvantaged and to support these students in their attainment and attitudes to secondary school science. Purpose This paper reports on a project designed to support students from disadvantaged backgrounds by pairing them with undergraduate mentors from a university throughout the final year of their compulsory science education in England (Year 11 – aged 15–16) at the end of which students take their public General Certificate of Secondary Education (GCSE) examinations. Sample The study, set up as a randomised control trial, involved 86 disadvantaged students – students from low-income families who are eligible for free school meals, or had been looked after for more than six months, or whose parent(s) are currently in the Armed Forces. Four schools were recruited. The schools were similar in terms of the proportion of free school meals, GCSE 5A*-C measures and value-added performance to reduce the likelihood of any effect being attributable to factors other than mentoring. Design and methods Students were recruited from four schools and were randomly assigned to experimental and control groups – half in each group. Experimental group students were mentored for one hour per week for 23 weeks up until their GCSE examinations with an intensive six-hour mentoring session just prior to those examinations. Data were collected from the Year 11’s mock and actual GCSE examination results as well as questionnaires from Year 11 and undergraduate mentors. Results It was found that mentored students did statistically better in terms of attainment in both mock and actual GCSEs examinations and also showed a statistically significantly greater improvement in their attitudes to science than un-mentored students. Conclusions These findings indicate the value of academically asymmetrical paired mentoring for disadvantaged students within science.