- Summary Objectives Burden of pneumococcal disease depends on the prevalence and invasive disease potential of serotypes. We aimed to estimate the invasive disease potential of serotypes in children under 5 years of age by combining data from different settings with routine immunisation with pneumococcal conjugate vaccines (PCV). Methods We conducted a systematic review, supplemented by unpublished data, to identify data on the frequency of pneumococcal serotypes in carriage and invasive pneumococcal disease (IPD). We estimated the invasive disease potential of serotypes as the ratio of IPD in relation to carriage (odds ratio and 95%CI) compared with 19A (reference serotype) by meta-analysis. We report results based on a random effects model for children aged 0–23, 24–29, and 0–59 months and by invasive clinical syndromes. Results In comparison with 19A, serotypes 1, 7F, and 12F had a significantly higher invasive disease potential in children aged 0–23 and 0–59 months for all IPD and clinical syndromes (OR>5). Several non-vaccine types (NVTs) (6C, 15A, 15BC, 16F, 23B, in these two age groups) had a lower invasive disease potential than 19A (OR 0•1−0•3). NVTs 8, 12F, 24F, and 33F were at the upper end of the invasiveness spectrum. Conclusions There is substantial variation among pneumococcal serotypes in their potential to cause IPD and disease presentation, which is influenced by age and time after PCV introduction. Surveillance of IPD and carriage is critical to understand the expected effectiveness of current PCVs (in the longer term) and guide the development of future vaccines.