We have analysed and compared the fine specificity and behavior in various immunoassays of 10 mouse monoclonal antibodies, from three independent laboratories, directed against IgA1, IgA2 or non-IgA2m(2). The following observations were made. (1) Although all of the monoclonal antibodies were specific for a particular IgA subclass or isoallotype in a radioimmunoassay, three of them were not specific when tested in indirect immunofluorescence on plasma cells derived from pokeweed-activated peripheral blood lymphocytes. In this highly sensitive system, contrary to direct immunofluorescence previously performed using formalin-fixed lymphoid tissue, the anti-IgA1 69.114 reacted with some of the IgA2 plasma cells, the anti-IgA2 DLDB7 reacted with some of the IgA1 plasma cells and the anti-IgA2 16.512 dimly reacted with all IgM plasma cells. (2) Among the eight anti-IgA subclass antibodies, seven were directed against the CH2 domain of IgA whereas the anti-IgA1 1-155-1 recognised an epitope destroyed by Streptococcus sanguis IgA1 protease and localised in the hinge region of IgA1. The two anti-isoallotype antibodies were directed against epitope(s) probably localised in the 65 C-terminal amino acid residues of the alpha-CH3 domain. All of the 10 antibodies were able to react with endogeneously produced surface IgA on B-cells. (3) Using monoclonal anti-IgA subclass antibodies in radioimmunoassay may be hazardous in the absence of knowledge of their affinity constants and of careful control experiments: some of the antibodies were not sensitive in radioimmunoassays designed to measure the serum titer of specific IgA1 and IgA2 antibodies. Moreover, major differences were observed between the different monoclonal reagents with respect to the influence of the size of IgA on a solid-phase sandwich radioimmunoassay. While three of the anti-IgA1 underestimated dimeric IgA relative to monomeric IgA, the fourth anti-IgA1 and all the anti-IgA2 overestimated dimeric IgA relative to monomeric IgA, by a factor sometimes close to 7.