BACKGROUND: Tuberculosis vaccine trials using disease as the primary endpoint are large, time consuming, and expensive. An earlier immunological measure of the protection against disease would accelerate tuberculosis vaccine development. We aimed to assess whether the effectiveness of the Bacillus Calmette-Guérin (BCG) vaccine for prevention of Mycobacterium tuberculosis infection was consistent with that for prevention of tuberculosis disease. METHODS: We conducted an individual participant data (IPD) meta-analysis on experimental and observational longitudinal studies before April 6, 2018, identified through systematic reviews, known to us through expert knowledge in the field, reporting on BCG vaccination status, M tuberculosis infection test (QuantiFERON IFN-γ release assay [IGRA] and tuberculin skin test [TST]), and tuberculosis incidence. Cohort studies were included only for countries with a mandatory neonatal BCG vaccination policy. Exclusion criteria were previous or current tuberculosis disease, HIV infection, tuberculosis preventive treatment usage, and for household contacts, a positive baseline IGRA or TST test and young children aged 0-2 years; for randomised controlled trials, TST results within 2 years after random assignation were excluded. We contacted the investigators of the identified studies to provide IPD. We compared the protective efficacy of the BCG vaccine against M tuberculosis infection with that against tuberculosis disease using mixed-effects, multivariable proportional hazards modelling, by study type, M tuberculosis infection test (IGRA and TST), cutoff for defining test positivity, age, sex, and latitude. FINDINGS: We identified 79 studies eligible for full screening and of these, IPD datasets from 14 studies were included in our analysis: 11 household contact studies (29 147 participants), two adolescent cohort studies (11 368 participants), and one randomised controlled trial (2963 participants). Among 28 188 participants we found no protection by the BCG vaccine against TST conversion regardless of cutoff in any type of study. Among 1491 household contacts, but not among 5644 adolescents, the BCG vaccine protected against QuantiFERON conversion at the primary cutoff of 0·7 IU/mL or more with the adjusted hazard ratio (0·65, 95% CI 0·51-0·82) being consistent with that for protection against disease (0·68, 0·18-2·59). Protection against QuantiFERON conversion at cutoff of 0·35 IU/mL or more (0·64, 0·51-0·81) was similar. INTERPRETATION: Protection from the BCG vaccination against M tuberculosis infection, measured as QuantiFERON conversion, is inconsistent across different groups. Among groups with recent household exposure, QuantiFERON conversion is consistent with protection against disease and could be evaluated as a proxy for disease in tuberculosis vaccine trials. We found that TST lacks value for prevention in phase 2b proof-of-concept trials. FUNDING: Bill & Melinda Gates Foundation.