Type 1 diabetes mellitus (T1DM) has been increasing in prevalence in the last decades and has become a global burden. Autoantibodies against human glutamate decarboxylase (GAD65) are among the first to be detected at the onset of T1DM. Diverse viruses have been proposed to be involved in the triggering of T1DM because of molecular mimicry, i.e., similarity between parts of some viral proteins and one or more epitopes of GAD65. However, the possibility that bacterial proteins might also be responsible for GAD65 mimicry has been seldom investigated. To date, many genomes of Streptococcus pneumoniae (the pneumococcus), a prominent human pathogen particularly prevalent among children and the elderly, have been sequenced. A dataset of more than 9000 pneumococcal genomes was mined and two different (albeit related) genes (gadA and gadB), presumably encoding two glutamate decarboxylases similar to GAD65, were found. The various gadASpn alleles were present only in serotype 3 pneumococci belonging to the global lineage GPSC83, although some homologs have also been discovered in two subspecies of Streptococcus constellatus (pharyngis and viborgensis), an isolate of the group B streptococci, and several strains of Lactobacillus delbrueckii. Besides, gadBSpn alleles are present in > 10% of the isolates in our dataset and represent 16 GPSCs with 123 sequence types and 20 different serotypes. Sequence analyses indicated that gadA- and gadB-like genes have been mobilized among different bacteria either by prophage(s) or by integrative and conjugative element(s), respectively. Substantial similarities appear to exist between the putative pneumococcal glutamate decarboxylases and well-known epitopes of GAD65. In this sense, the use of broader pneumococcal conjugate vaccines such as PCV20 would prevent the majority of serotypes expressing those genes that might potentially contribute to T1DM. These results deserve upcoming studies on the possible involvement of S. pneumoniae in the etiopathogenesis and clinical onset of T1DM.