BACKGROUND: Trials of BCG vaccination to prevent or reduce severity of COVID-19 are taking place in adults, some of whom have been previously vaccinated, but evidence of the beneficial, non-specific effects of BCG come largely from data on mortality in infants and young children, and from in-vitro and animal studies, after a first BCG vaccination. We assess all-cause mortality following a large BCG revaccination trial in Malawi. METHODS: The Karonga Prevention trial was a population-based, double-blind, randomised controlled in Karonga District, northern Malawi, that enrolled participants between January, 1986, and November, 1989. The trial compared BCG (Glaxo-strain) revaccination versus placebo to prevent tuberculosis and leprosy. 46 889 individuals aged 3 months to 75 years were randomly assigned to receive BCG revaccination (n=23 528) or placebo (n=23 361). Here we report mortality since vaccination as recorded during active follow-up in northern areas of the district in 1991-94, and in a demographic surveillance follow-up in the southern area in 2002-18. 7389 individuals who received BCG (n=3746) or placebo (n=3643) lived in the northern follow-up areas, and 5616 individuals who received BCG (n=2798) or placebo (n=2818) lived in the southern area. Year of death or leaving the area were recorded for those not found. We used survival analysis to estimate all-cause mortality. FINDINGS: Follow-up information was available for 3709 (99·0%) BCG recipients and 3612 (99·1%) placebo recipients in the northern areas, and 2449 (87·5%) BCG recipients and 2413 (85·6%) placebo recipients in the southern area. There was no difference in mortality between the BCG and placebo groups in either area, overall or by age group or sex. In the northern area, there were 129 deaths per 19 694 person-years at risk in the BCG group (6·6 deaths per 1000 person-years at risk [95% CI 5·5-7·8]) versus 133 deaths per 19 111 person-years at risk in the placebo group (7·0 deaths per 1000 person-years at risk [95% CI 5·9-8·2]; HR 0·94 [95% CI 0·74-1·20]; p=0·62). In the southern area, there were 241 deaths per 38 399 person-years at risk in the BCG group (6·3 deaths per 1000 person-years at risk [95% CI 5·5-7·1]) versus 230 deaths per 38 676 person-years at risk in the placebo group (5·9 deaths per 1000 person-years at risk [95% CI 5·2-6·8]; HR 1·06 [95% CI 0·88-1·27]; p=0·54). INTERPRETATION: We found little evidence of any beneficial effect of BCG revaccination on all-cause mortality. The high proportion of deaths attributable to non-infectious causes beyond infancy, and the long time interval since BCG for most deaths, might obscure any benefits. FUNDING: British Leprosy Relief Association (LEPRA); Wellcome Trust.