BACKGROUND: Surgical site infections (SSIs) are the most frequent health-care-associated infections in developing countries. Specific prevention measures are highly effective, but are often poorly implemented. We aimed to establish the effect of a multimodal intervention on SSIs in Africa. METHODS: We did a before-after cohort study, between July 1, 2013, and Dec 31, 2015, at five African hospitals. The multimodal intervention consisted of the implementation or strengthening of multiple SSI prevention measures, combined with an adaptive approach aimed at the improvement of teamwork and the safety climate. The primary outcome was the first occurrence of SSI, and the secondary outcome was death within 30 days post surgery. Data on adherence to SSI prevention measures were prospectively collected. The intervention effect on SSI risk and death within 30 days post surgery was assessed in a mixed-effects logistic regression model, after adjustment for key confounders. FINDINGS: Four hospitals completed the baseline and follow-up; three provided suitable (ie, sufficient number and quality) data for the sustainability period. 4322 operations were followed up (1604 at baseline, 1827 at follow-up, and 891 in the sustainability period). SSI cumulative incidence significantly decreased post intervention, from 8·0% (95% CI 6·8-9·5; n=129) to 3·8% (3·0-4·8; n=70; p<0·0001), and this decrease persisted in the sustainability period (3·9%, 2·8-5·4; n=35). A substantial improvement in compliance with prevention measures was consistently observed in the follow-up and sustainability periods. The likelihood of SSI during follow-up was significantly lower than pre-intervention (odds ratio [OR] 0·40, 95% CI 0·29-0·54; p<0·0001), but the likelihood of death was not significantly reduced (0·72, 0·42-1·24; p=0·2360). INTERPRETATION: Implementation of our intervention is feasible in African hospitals. Improvement was observed across all perioperative prevention practices. A significant effect on the overall SSI risk was observed, but with some heterogeneity between sites. Further large-scale experimental studies are needed to confirm these results and to improve the sustainability and long-term effect of such complex programmes. FUNDING: US Agency for Healthcare Research and Quality, WHO.