This doctoral thesis presents the findings of a mixed methods case study of Learning Skills, a new approach to Learning to Learn that was developed and implemented at a secondary school in the south of England between 2010 and 2014, and evaluated using data collected between 2009 and 2017. Learning to Learn is a field of educational theory and practice that aims to help young people get better at learning by focusing on the processes of learning (the how as well as the what), and by enabling them to take ownership over aspects of their own learning through activities such as goal setting, self-monitoring and structured reflection. The field has developed significantly throughout the last 40 years, with a number of approaches having been implemented on a large scale in the UK. Research into metacognition and self-regulation suggests that Learning to Learn programmes should help boost academic attainment. To date however, large-scale evaluations have found mixed results, with no clear impact on academic attainment. Using an intervention design used widely in medicine and other fields, Learning Skills reconceptualises Learning to Learn as a ‘complex intervention’ comprised of multiple areas of evidence-informed practice. The rationale for complex interventions is that the marginal gains emerging from any individual avenue of practice stack up and interact to yield a larger effect size overall. The Learning Skills programme, which started as a year seven taught course and developed into a whole-school approach to teaching and learning, focuses centrally on three key concepts: metacognition, self-regulation and oracy. This evaluation of Learning Skills incorporates eight strands of data collection and analysis over an eight-year period, using the previous year group at the same school as a control group. These include baseline measures; attitude to learning scores; psychometric questionnaires; a language of learning evaluation; reflective learning journals; student interviews; teacher interviews; and student attainment across all subjects in years nine and 11. The primary outcome analysis – student attainment across all subject areas at three and five years – found that Learning Skills cohort one achieved significantly higher grades than the control cohort, with accelerated gains among young people from economically disadvantaged backgrounds. Secondary data analysis incorporating a range of qualitative and quantitative methods indicates a causal relationship between Learning Skills and academic attainment. As well as evaluating the impact of a new and promising approach to Learning to Learn, this study generates new knowledge about the implementation and evaluation of complex interventions in education.