Background: Risk of mortality following surgery in patients across Africa is twice as high as the global average. Most of these deaths occur on hospital wards after the surgery itself. We aimed to assess whether enhanced postoperative surveillance of adult surgical patients at high risk of postoperative morbidity or mortality in Africa could reduce 30-day in-hospital mortality. Methods: We did a two-arm, open-label, cluster-randomised trial of hospitals (clusters) across Africa. Hospitals were eligible if they provided surgery with an overnight postoperative admission. Hospitals were randomly assigned through minimisation in recruitment blocks (1:1) to provide patients with either a package of enhanced postoperative surveillance interventions (admitting the patient to higher care ward, increasing the frequency of postoperative nursing observations, assigning the patient to a bed in view of the nursing station, allowing family members to stay in the ward, and placing a postoperative surveillance guide at the bedside) for those at high risk (ie, with African Surgical Outcomes Study Surgical Risk Calculator scores ≥10) and usual care for those at low risk (intervention group), or for all patients to receive usual postoperative care (control group). Health-care providers and participants were not masked, but data assessors were. The primary outcome was 30-day in-hospital mortality of patients at low and high risk, measured at the participant level. All analyses were done as allocated (by cluster) in all patients with available data. This trial is registered with ClinicalTrials.gov, NCT03853824. Findings: Between May 3, 2019, and July 27, 2020, 594 eligible hospitals indicated a desire to participate across 33 African countries; 332 (56%) were able to recruit participants and were included in analyses. We allocated 160 hospitals (13 275 patients) to provide enhanced postoperative surveillance and 172 hospitals (15 617 patients) to provide standard care. The mean age of participants was 37·1 years (SD 15·5) and 20 039 (69·4%) of 28 892 patients were women. 30-day in-hospital mortality occurred in 169 (1·3%) of 12 970 patients with mortality data in the intervention group and in 193 (1·3%) of 15 242 patients with mortality data in the control group (relative risk 0·96, 95% CI 0·69–1·33; p=0·79). 45 (0·2%) of 22 031 patients at low risk and 309 (5·6%) of 5500 patients at high risk died. No harms associated with either intervention were reported. Interpretation: This intervention package did not decrease 30-day in-hospital mortality among surgical patients in Africa at high risk of postoperative morbidity or mortality. Further research is needed to develop interventions that prevent death from surgical complications in resource-limited hospitals across Africa. Funding: Bill & Melinda Gates Foundation and the World Federation of Societies of Anaesthesiologists. Translations: For the Arabic, French and Portuguese translations of the abstract see Supplementary Materials section.