With the advent of the genomic era, identification of bacterial factors involved in virulence is a different challenge. Given the vast amount of information available on toxins, in terms of sequence and 3D structure, and thanks to the growing number of sequenced bacterial genomes, it is possible to proceed by homology criteria to predict novel toxins in different microorganisms. ADP-ribosyltransferases constitute a class of functionally conserved enzymes, which display toxic activity in a variety of bacterial pathogens. Since these proteins play a key role in pathogenesis, they have been extensively characterized and successfully used as vaccine components and mucosal adjuvants. Therefore, the application of in silico analyses to identify novel members of this class of enzymes represents an important challenge in the genomic era. To address this subject, different groups have recently pursued homology-based procedures to screen bacterial genomes for novel, yet undiscovered ADP-ribosyltransferases (ADPRTs) and have identified more than twenty novel ADPRTs in Gram-positive and Gram-negative bacteria. We have developed a novel pattern-based computational approach, which, flanked by secondary structure prediction tools, has allowed the identification of previously unrecognised putative ADPRTs. One of them, identified in Neisseria meningitidis has been extensively characterized and shown to possess the predicted enzymatic activity, suggesting a possible role of this protein in the virulence of Meningococcus.