Summary Objectives: To report the lessons learned from eight years of feeding back routinely collected cardiovascular data in an educational context Methods: There are distinct educational and technical components. The educational component provides peer-led learning opportunities based on comparative analysis of quality of care, as represented in computer records. The technical part ensures that relevant evidence-based audit criteria are identified; an appropriate dataset is extracted and processed to facilitate quality improvement. Anonymised data are used to provide inter-practice comparisons, with lists of identifiable patients who need interventions left in individual practices. Results: The progressive improvement in cholesterol management in ischaemic heart disease (IHD) is used as an exemplar of the changes achieved. Over three iterations of the cardiovascular programme the standardised prevalence of IHD recorded in GP computer systems rose from 3.8% to 4.0%. Cholesterol recording rose from 47.6% to 89.0%; and the mean cholesterol level fell from 5.18 to 4.67 mmol/L; while statin prescribing rose from 46% to 57% to 68%. The atrial fibrillation, heart failure and renal programmes (more people with chronic kidney disease go on to die from cardiovascular cause than from end-stage renal disease) are used to demonstrate the range of cardiovascular interventions amenable to this approach. Conclusions: Technical progress has meant that larger datasets can be extracted and processed. Feedback of routinely collected data in an educational context is acceptable to practitioners and results in quality improvement. Further research is needed to assess its utility as a strategy and cost-effectiveness compared with other methods.