Summary Background Despite widespread use of intrapartum antibiotic prophylaxis, group B streptococcus remains a leading cause of morbidity and mortality in infants in Europe, the Americas, and Australia. However, estimates of disease burden in many countries outside of these regions is not available. We aimed to examine the current global burden of invasive disease and the serotype distribution of group B streptococcus isolates. Methods We searched Medline, Embase, and Wholis databases for studies on invasive early-onset (day 0–6) and late-onset (day 7–89) group B streptococcal disease. Eligible studies were those that described incidence, deaths, or serotypes. We also reviewed reference lists and contacted experts to seek unpublished data and data missed by our search. Random effects meta-analysis was used to pool data. Findings 74 studies met the inclusion criteria; 56 studies reported incidence, 29 case fatality, and 19 serotype distribution. An additional search for studies that reported serotype distribution from Jan 1, 1980, yielded a total of 38 articles. Only five low-income countries were represented in the review and contributed 5% weight to the meta-analysis. 47 (69%) studies reported use of any intrapartum antibiotic prophylaxis. Substantial heterogeneity existed between studies. Mean incidence of group B streptococcus in infants aged 0–89 days was 0·53 per 1000 livebirths (95% CI 0·44–0·62) and the mean case fatality ratio was 9·6% (95% CI 7·5–11·8). Incidence of early-onset group B streptococcus (0·43 per 1000 livebirths 95% CI 0·37–0·49) and case fatality (12·1%, 6·2–18·3) were two-times higher than late-onset disease. Serotype III (48·9%) was the most frequently identified serotype in all regions with available data followed by serotypes Ia (22·9%), Ib (7·0%), II (6·2%), and V (9·1%). Studies that reported use of any intrapartum antibiotic prophylaxis were associated with lower incidence of early-onset group B streptococcus (0·23 per 1000 livebirths 95% CI 0·13–0·59) than studies in which patients did not use prophylaxis (0·75 per 1000 livebirths 0·58–0·89). Interpretation More high-quality studies are needed to accurately estimate the global burden of group B streptococcus, especially in low-income countries. A conjugate vaccine incorporating five serotypes (Ia, Ib, II, III, V) could prevent most global group B streptococcal disease. Funding Child Epidemiology Reference Group (CHERG), WHO.