CVID are a group of heterogeneous conditions characterized by reduced immunoglobulin levels and absent or poor antibody responses. The latter diagnostic criterion has not been clearly defined leading to a highly variable number and type of immunizations performed among centres. Specific antibody responses cannot be evaluated in patients who are already on immunoglobulin replacement, due to the interference of passively administered IgG. Classification schemes are based on cellular phenotyping and offer instruments for recognition of patients at risk for CVID-associated clinical conditions, but do not take advantage of the possibility to evaluate in vivo antibody responses as a prognostic marker for infectious complications. We immunized 91 CVID patients with a 23-valent pneumococcal polysaccharide vaccine (Pneumovax®) and measured the IgM and IgA to single pneumococcal polysaccharides before vaccination and 4 weeks later. Results were compared with those obtained using a new IgM and IgA anti-pneumococcal polysaccharides 23-valent assay (PC23). We demonstrated that the IgM/IgA response to PC23 allows stratifying CVID patients into groups with different risk to experience pneumonia and to develop bronchiectasis. Immunological IgM/IgA responders had the lowest risk for pneumonia (0%) and bronchiectasis (1.2%), while non responders had the highest risk (37% and 41.5% respectively) and IgM-only responders had an intermediate risk (8.8% and 8% respectively). The antibody response correlated with the frequency of IgMpos and switched memory B cells. The IgM and IgA PC23 assay represents a valuable prognostic tool for CVID patients and allows investigating the residual antibody production capacity, even in patients on substitutive immunoglobulin replacement.