Minimising suffering is an ethical and legal requirement in animal research. This is particularly relevant for research on animal models of sepsis and septic shock, which show rapid progression towards severe stages and death. Specific and reliable criteria signalling non-recovery points can be used as humane endpoints, beyond which a study cannot be allowed to progress, thus preventing avoidable suffering. Body temperature is a key indicator for assessing animal health and welfare and has been suggested to have potential for monitoring the status of mouse models of sepsis. In this study, we monitored temperature variations using contactless methods – thermal imaging and subcutaneously implanted PIT tags – in a surgical model of sepsis by caecal ligation and puncture (CLP). We monitored body temperature variation following mid-grade CLP, high-grade CLP and sham surgery. All mice (Mus musculus) were monitored four times per day in the high-grade CLP model and three times per day in the mid-grade CLP model by both PIT tag readout and infrared thermography for ten days post-surgery, or until animals reached a predefined humane endpoint. Thermal data were compared with the clinical score and weight loss threshold used at our facility. Mean body surface temperature (MBST) assessed by thermal imaging and subcutaneous temperature (SCT) measured by PIT tags correlated, albeit not strongly. Moreover, while MBST does not appear to be a reliable predictor of non-recovery stages, SCT showed promise in this regard, even surpassing the widely used weight loss criterion, particularly for the high-grade CLP model of induced sepsis.