WHO recommends delaying measles vaccination (MV) until maternal antibody has waned. However, early MV may improve child survival by reducing mortality from conditions other than measles infection. We tested whether early MV improves child survival compared with later MV. We found 43 studies comparing measles-vaccinated and measles-unvaccinated children; however, only 16 studies had specific information that MV had been provided at 4-13 months of age, many before 9 months of age. In the 10 best studies (4 randomized trials and 6 observational studies) control children did not receive MV during follow-up. In eight of these studies the vaccine efficacy against death (VED) was 60% or more. In four studies with information on MV provided both before and after 12 months of age, the all-cause mortality reduction was significantly larger for children vaccinated in infancy (VED=74%; 95% CI 51-86%) than for children vaccinated after 12 months of age (VED=29%; CI 8-46%). Prevention of measles explained little of the reduction in mortality. In five studies with information on measles infection, VED was 67% (51-78%) and when measles deaths were excluded, VED was only reduced to 65% (47-77%). One natural experiment compared MV at 4-8 months versus MV at 9-11 months of age and found significantly lower all-cause mortality with early vaccination, the difference being 39% (8-60%). Child mortality may be reduced if MV is given earlier than currently recommended by international organizations.