Microbes employ a large diversity of toxins to kill competing microbes or eukaryotic host cells. Polymorphic toxins are a class of protein toxins abundant in Nature and secreted through various secretion systems. Polymorphic toxins have a modular protein architecture with a toxin domain found at the protein C-terminus. The discovery of microbial toxins is important to improve our understanding of microbial ecology and infectious diseases. Here, we developed a computational approach to discover novel toxin domains of polymorphic toxins and applied it to 105,438 microbial genomes. We validated nine short novel toxins (“PTs”) that lead to bacterial cell death; one of them also kills yeast. For four toxins, we also identified a cognate immunity gene (“PIM”) that protects the toxin producing cell from toxicity. Additionally, we found that the novel PTs are encoded by ~2.2% of the sequenced bacteria, including numerous Gram-negative and -positive pathogens. We explored two avenues to determine the mechanism of action of these nine PTs. First, we used fluorescence microscopy to observe severe changes in cell size, membrane and chromosome morphology upon expression of the novel toxins. We then identified putative catalytic sites of these toxins, which, upon mutation, abolished their harmful activities. Thus, using this unique hybrid computational-experimental pipeline, we were able to expand the microbial toxins repertoire significantly. These new potent toxins likely play essential roles in inter-microbial competition in Nature and can be utilized in various biotechnological and clinical applications.