BACKGROUND: The available data on antimicrobial stewardship programmes in Africa are scarce. The aims of this study were to assess the implementation of an antimicrobial stewardship programme in a setting with limited infectious disease resources. METHODS: We implemented a pharmacist-driven, prospective audit and feedback strategy for antimicrobial stewardship on the basis of a range of improvement science and behavioural principles across a diverse group of urban and rural private hospitals in South Africa. The study had a pre-implementation phase, during which a survey of baseline stewardship activities was done. Thereafter, a stepwise implementation phase was initiated directed towards auditing process measures to reduce consumption of antibiotics (prolonged duration, multiple antibiotics, and redundant antibiotic coverage), followed by a post-implementation phase once the model was embedded in each hospital. The effect on consumption was assessed with the WHO index of defined daily doses per 100 patient-days, and the primary outcome (change in antibiotic consumption between phases) was assessed with a linear mixed-effects regression model. FINDINGS: We implemented and assessed the antimicrobial stewardship programme between Oct 1, 2009, and Sept 30, 2014. 116 662 patients receiving antibiotics at 47 hospitals during 104 weeks of standardised measurement and feedback, were reviewed, with 7934 interventions by pharmacists recorded for the five targeted measures, suggesting that almost one in 15 prescriptions required intervention. 3116 (39%) of 7934 pharmacist interventions were of an excessive duration. The antimicrobial stewardship programme led to a reduction in mean antibiotic defined daily doses per 100 patient-days from 101·38 (95% CI 93·05-109·72) in the pre-implementation phase to 83·04 (74·87-91·22) in the post-implementation phase (p<0·0001). INTERPRETATION: Health-care facilities with limited infectious diseases expertise can achieve substantial returns through pharmacist-led antimicrobial stewardship programmes and by focusing on basic interventions. FUNDING: None.