The sandwich microarray immunoassay (SMI) is a powerful technique for the analysis and characterization of environmental samples, from the identification of microorganisms to specific bioanalytes. As the number of antibodies increases, however, unspecific binding and cross-reactivity can become a problem. To cope with such difficulties, we present here the concept of antibody graph associated to a sandwich antibody microarray. Antibody graphs give valuable information about the antibody cross-reactivity network and all the players involved in the sandwich format: capturing and tracer antibodies, the antigenic sample and the degree of cross-reactivity between antibodies. Making use of the information contained in the antibody graph, we have developed a deconvolution method that disentangles the antibody cross-reactivity events and gives qualitative information about the composition of the experimental sample under study. We have validated the method by using a 66 antibody-containing microarray to describe known antigenic mixtures as well as natural environmental samples characterized by 16S-RNA gene phylogenetic analysis. The application of our antibody graph and deconvolution method allowed us to discriminate between true specific antigen–antibody reactions and spurious signals on a microarray designed for environmental monitoring.