Antimicrobial resistance represents one of the main threats to global health due to some pathogens being extremely resistant to existing antibiotics. In this scenario, it is necessary for new drugs to become promising. To this end, in vivo nonclinical trials are an important step in the development of new drugs due to their relatively low cost, the possibility of mimicking pathological conditions in living organisms, and the fact that they provide relevant data on toxicity and antibacterial activity. However, there is no homogeneity in the studies regarding the techniques and protocols used to achieve the desired clinical conditions. This is a narrative review, the objective of which is to evaluate which models are used to induce sepsis in mice. The research was carried out using the PubMed and Virtual Health Library databases, with the descriptors “models of bacterial infection”, “mus musculus”, “sepsis”, and “in vivo”, using the AND and OR connectors. Twenty-seven articles were selected, of which 40% used the intraperitoneal technique—with the species Escherichia coli and Staphylococcus aureus being the most commonly used—33% were performed with cecal puncture ligation (CPL), and in the other 26%, the administration route was intravenous, preferably via the caudal route. In most studies, the final disease induced was sepsis, differing in relation to the focus, which varied between urinary, abdominal, and pulmonary. The main organs removed were the spleen, lungs, liver, and kidneys, which were subjected to histopathological examination and bacterial count. Thus, it is concluded that the most commonly used models are obtained via the intraperitoneal administration route and CPL and that there is a wide range of protocols used for confirmation.