Immunocompromised patients (ICPs) have a higher risk of developing severe forms of COVID-19 and experience a higher burden of complications and mortality than the general population. However, recent studies have suggested that the antibody response to SARS-CoV-2 mRNA vaccines could be highly variable among different ICPs. Using a collaborative, monocentric, prospective cohort study, we assessed anti-SARS-CoV-2 spike protein antibody titers following two and three doses of mRNA vaccines in four groups of ICPs (cancer [n= 232]: hematopoietic stem cell transplant [HSCT;n= 126] patients; people living with HIV [PLWH;n= 131]; and lung transplant [LT;n= 39] recipients) treated at Geneva University Hospitals; and healthy individuals (n= 49). After primo-vaccination, the highest anti-S antibody geometric mean titer (IU/mL) was observed in healthy individuals (2417 IU/mL [95% CI: 2327-2500]), the PLWH group (2024 IU/mL [95% CI:1854-2209]) and patients with cancer (840 IU/mL [95% CI: 625-1129]), whereas patients in the HSCT and LT groups had weaker antibody responses (198 IU/mL [95% CI: 108-361] and 7.3 IU/mL [95% CI: 2.5-22]). The booster dose conferred a high antibody response after 1 month in both PLWH (2500 IU/mL) and cancer patients (2386 IU/mL [95% CI: 2182-2500]), a moderate response in HSCT patients (521 IU/mL [95% CI: 306-885]) and a poor response in LT recipients (84 IU/mL [95% CI: 18-389]). Contemporary treatment with immunosuppressive drugs used in transplantation or chemotherapy was associated with a poor response to vaccination. Our findings confirmed the heterogeneity of the humoral response after mRNA vaccines among different ICPs and the need for personalized recommendations for each of these different groups.