BACKGROUND: The risk of tuberculosis disease after recent exposure is greatest before age 5 years; however, the mechanisms explaining this increased risk are not well elucidated. Acquisition of viral infections, such as cytomegalovirus, in early life might modulate the immune system. We aimed to evaluate the acquisition of cytomegalovirus infection in infancy and the development of tuberculosis disease in children. METHODS: In this prospective, birth cohort study we enrolled pregnant women who were between 20 and 28 weeks of gestation attending antenatal care in Paarl, a periurban setting outside of Cape Town, South Africa. Participants were recruited from two clinics (TC Newman and Mbekweni). Infants were given Bacillus Calmette-Guérin vaccination at birth as per national policy. Nasopharyngeal swabs for cytomegalovirus detection using qPCR were done for infants at birth, age 3 and 6 weeks, and age 3, 6, 12, and 24 months. Children were prospectively followed up for tuberculosis disease until age 9 years using tuberculin skin testing, radiographic examinations, GeneXpert, and sputum testing. Tuberculin skin tests were done at the 6-month visit and then at age 12, 24, 36, 48, and 60 months, and at the time of lower respiratory tract infection. We compared tuberculosis disease incidence after age 1 year or after age 6 months in children with and without cytomegalovirus infection using Cox regression and hazard ratios (HRs) with 95% CIs. FINDINGS: Between March 5, 2012, and March 31, 2015, 1225 pregnant women were recruited and enrolled in the birth cohort. 88 (7%) women were excluded because of loss to antenatal follow-up or pregnancy losses. Of 1143 livebirths, 68 (6%) mother-infant pairs were excluded. In total, 963 children were serially tested for cytomegalovirus (7186 cytomegalovirus measurements taken; median six tests per child, IQR 2-11). The prevalence of congenital cytomegalovirus at age younger than 3 weeks was 2% (18 of 816). Cytomegalovirus positivity increased continuously with age from 3% (27 of 825) by age 6 weeks to 21% (183 of 882) by 3 months, 35% (315 of 909) by 6 months, and 42% (390 of 933) by 12 months. Mother-infant pairs were followed up for a median of 6·9 years (IQR 6·0-7·8). The risk of tuberculosis disease in children after age 1 year was higher in those with cytomegalovirus infection by age 6 weeks (adjusted HR 4·1, 95% CI 1·2-13·8; p=0·022), 3 months (2·8, 1·4-5·8; p=0·0040), 6 months (3·6, 1·7-7·3; p<0·0001), 12 months (3·2, 1·6-6·4; p=0·0010), and 24 months (4·2, 2·0-8·8; p<0·0001). The risk of microbiologically confirmed tuberculosis disease was also higher among children acquiring cytomegalovirus infection before age 3 months (adjusted HR 3·2, 95% CI 1·0-10·6; p=0·048), 6 months (3·9, 1·2-13·0; p=0·027), 12 months (4·4, 1·2-16·3; p=0·027), and 24 months (6·1, 1·3-27·9; p=0·020). In children older than 1 year, the risk of tuberculosis disease was consistently greater in those with high cytomegalovirus loads than in those with low cytomegalovirus loads that were acquired before age 3 months (adjusted HR 2·0 vs 3·7; ptrend=0·0020; both groups compared with cytomegalovirus negative reference) and before age 12 months (2·7 vs 3·7; ptrend=0·0009). INTERPRETATION: Infants that acquire cytomegalovirus in the first year of life are at high risk of subsequently developing tuberculosis disease. Efforts to prevent tuberculosis in early childhood in high-burden countries might need to deter or delay acquisition of cytomegalovirus perinatally or in the first months of life. FUNDING: Bill & Melinda Gates Foundation, MRC South Africa, National Research Foundation South Africa, and Wellcome Trust.