Death after surgery is devasting for patients, families, and communities, but remains common in low-income and middle-income countries (LMICs). We aimed to use high-quality data from an existing global randomised trial to describe the causes and mechanisms of postoperative mortality in LMICs. To do so, we developed a novel framework, learning from both existing classification systems and emerging insights during data analysis. This study was a preplanned secondary analysis of the FALCON trial in 54 hospitals across seven LMICs (Benin, Ghana, India, Mexico, Nigeria, Rwanda, and South Africa). FALCON was a pragmatic, 2 × 2 factorial, randomised controlled trial that compared the effectiveness of two types of interventions for skin preparation (10% aqueous povidone-iodine vs 2% alcoholic chlorhexidine) and sutures (triclosan-coated vs uncoated). Patients who did not have surgery or were lost to follow-up were excluded (n=231). The primary outcomes of the present analysis were the mechanism and cause of death within 30-days of surgery, determined using a modified verbal autopsy strategy from serious adverse event reports. Factors associated with mortality were explored in a mixed-effects Cox proportional hazards model. The FALCON trial is registered with ClinicalTrials.gov, NCT03700749. This preplanned secondary analysis of the FALCON trial included 5558 patients who underwent abdominal surgery, of whom 4248 (76·4%) patients underwent surgery in tertiary, referral centres and 1310 (23·6%) underwent surgery in primary referral (ie, district or rural) hospitals. 3704 (66·7%) of 5558 surgeries were emergent. 306 (5·5%) of 5558 patients died within 30 days of surgery. 226 (74%) of 306 deaths were due to circulatory system failure, which included 173 (57%) deaths from sepsis and 29 (9%) deaths from hypovolaemic shock including bleeding. 47 (15%) deaths were due to respiratory failure. 60 (20%) of 306 patients died without a clear cause of death: 45 (15%) patients died with sepsis of unknown origin and 15 (5%) patients died of an unknown cause. 46 (15%) of 306 patients died within 24 h, 111 (36%) between 24 h and 72 h, 57 (19%) between >72 h and 168 h, and 92 (30%) more than 1 week after surgery. 248 (81%) of 306 patients died in hospital and 58 (19%) patients died out of hospital. The adjusted Cox regression model identified age (hazard ratio 1·01, 95% CI 1·01-1·02; p<0·0001), ASA grade III-V (4·93, 3·45-7·03; p<0·0001), presence of diabetes (1·47, 1·04-2·41; p=0·033), being an ex-smoker (1·59, 1·10-2·30; p=0·013), emergency surgery (2·08, 1·45-2·98; p<0·0001), cancer (1·98, 1·42-2·76; p<0·0001), and major surgery (3·94, 2·30-6·75; p<0·0001) as risk factors for postoperative mortality INTERPRETATION: Circulatory failure leads to most deaths after abdominal surgery, with sepsis accounting for almost two-thirds. Variability in timing of death highlights opportunities to intervene throughout the perioperative pathway, including after hospital discharge. A high proportion of patients without a clear cause of death reflects the need to improve capacity to rescue and cure by strengthening perioperative systems. National Institute for Health and Care Research Global Health Research Unit.