Background: The duration of protection offered by rotavirus vaccines varies across the world, and this variation is important to understanding and predicting the effects of the vaccines. There is now a large body of evidence on the efficacy of live oral rotavirus vaccines in different settings, but these data have never been synthesised to obtain robust estimates of efficacy by duration of follow-up. Our aim is to estimate the efficacy of live oral rotavirus vaccines at each point during follow-up and by mortality stratum. Methods: In our meta-regression study, we identified all randomised controlled trials of rotavirus vaccines published until April 4, 2018, using the results of a Cochrane systematic review, and cross checked these studies against those identified by another systematic review. We excluded trials that were based on special populations, trials without an infant schedule, and trials without clear reporting of numbers of enrolled infants and events in different periods of follow-up. For all reported periods of follow-up, we extracted the mean duration of follow-up (time since administration of the final dose of rotavirus vaccination), the number of enrolled infants, and case counts for rotavirus-positive severe gastroenteritis in both non-vaccinated and vaccinated groups. We used a Bayesian hierarchical Poisson meta-regression model to estimate the pooled cumulative vaccine efficacy (VE) and its waning with time for three mortality strata. We then converted these VE estimates into instantaneous VE (iVE). Findings: In settings with low mortality (15 observations), iVE pooled for infant schedules of Rotarix and RotaTeq was 98% (95% credibility interval 93–100) 2 weeks following the final dose of vaccination and 94% (87–98) after 12 months. In medium-mortality settings (11 observations), equivalent estimates were 82% (74–92) after 2 weeks and 77% (67–84) after 12 months. In settings with high mortality (24 observations), there were five different vaccines with observation points for infant schedules. The pooled iVE was 66% (48–81) after 2 weeks of follow-up and 44% (27–59) after 12 months. Interpretation: Rotavirus vaccine efficacy is lower and wanes more rapidly in high-mortality settings than in low-mortality settings, but the earlier peak age of disease in high-mortality settings means that live oral rotavirus vaccines are still likely to provide substantial benefit. Funding: Bill & Melinda Gates Foundation.