Background Matching donor red blood cells based on recipient antigens prevents alloimmunisation. Knowledge about the immunogenicity of red-blood-cell antigens can help optimise risk-adapted matching strategies. We set out to assess the immunogenicity of red-blood-cell antigens. Methods In an incident new-user cohort of previously non-transfused, non-alloimmunised white patients receiving non-extended matched red-blood-cell transfusions in six Dutch hospitals between 2006 and 2013, we determined the cumulative number of mismatched red-blood-cell units per patient. We used multiple imputation to address missing antigen data. Using Kaplan-Meier analysis, we estimated cumulative alloimmunisation incidences per mismatched antigen dose as a measure of immunogenicity. Findings Of 54 347 patients assessed, 21 512 were included in our study. Alloantibodies occurred in 474 (2·2%) of all transfused patients, with cumulative alloimmunisation incidences increasing up to 7·7% (95% CI 4·9–11·2) after 40 units received. The antigens C, c, E, K, and Jka were responsible for 78% of all alloimmunisations in our cohort. K, E, and Cw were the most immunogenic antigens (cumulative immunisation incidences after 2 mismatched units of 2·3% [95% CI 1·0–4·8] for K, 1·5% [0·6–3·0] for E, and 1·2% [0·0–10·8] for Cw). These antigens were 8·7 times (for K), 5·4 times (for E), and 4·6 times (for Cw) as immunogenic as Fya. The next most immunogenic antigens were, in order, e (1·9 times as immunogenic as Fya), Jka (1·9 times), and c (1·6 times). Interpretation Red-blood-cell antigens vary in their potency to evoke a humoral immune response. Our findings highlight that donor-recipient red-blood-cell matching strategies will be most efficient when primarily focusing on prevention of C, c, E, K, and Jka alloimmunisation. Matching for Fya is of lower clinical relevance. Variations of antigen frequencies determined by ethnic background prevent extrapolating these conclusions to non-white populations. Funding None.