Recent cohort studies showed differences in inhibitor incidence in previously untreated patients (PUPs) with haemophilia A treated with recombinant factor VIII (rFVIII) concentrates. We carried out a systematic literature search and meta-analysis for all randomized clinical trials and observational studies published from 1 January 1988 to 31 August 2015, to assess the incidence of inhibitor development and the relationship with rFVIII product used in PUPs and minimally treated patients (MTPs, ≤5 previous exposure days), with severe haemophilia. The primary outcome measure was development of all inhibitors and high-titre inhibitors. We computed pooled meta-analytic estimates according to the rFVIII product used with the inverse-variance method, assuming a fixed, or a random-effects model if significant between-studies heterogeneity was present. Out of 781 articles, 16 published between 1998 and 2015 were included in the meta-analysis, involving a total of 2094 haemophilia A PUPs or MTPs. The pooled estimate of all inhibitors was 0.27 (95% confidence interval 0.23–0.31). No significant difference in pooled inhibitor incidence across products was found (P = 0.72). Meta-analysis of studies reporting inhibitor hazard ratios with different rFVIII products, adjusted to different risk factors, showed that PUPs/MTPs treated with Advate had a pooled inhibitor hazard ratio estimate of 0.63 (95% confidence interval 0.48–0.83) as compared with Kogenate FS. The overall inhibitor incidence in PUPs/MTPs was 27%. Differences between products were found considering hazard ratios in which potential confounders were taken into account.