Controlled Human Infection Models (CHIMs) are a type of clinical trial involving deliberately exposing human volunteers to an infectious agent. Compared to studies of natural infection, CHIMs offers distinctive benefits, from the ability to study presymptomatic infection to a direct assessment of the efficacy of vaccines and therapeutics in a shorter time and involving fewer participants. Although the CHIMs do not fundamentally differ from other early phase clinical trials, they raise a unique set of ethical considerations. Modern CHIMs have been limited to studies which pose very low risk of participants experiencing serious adverse events, and CHIMs which lie on the upper end of risk remain controversial. Setting risk thresholds for more hazardous trials has become a topic of debate leading to various attempts to develop a model to ethically evaluate these risks. While there are now well-accepted practices around established pathogens, future pandemics may involve more extreme threats to public health and CHIMs may have an important emergency rescue function, with some groups willing to undertake significant risks. We must ask, what risk of death and serious morbidity should future CHIMs permit in the face of such challenges? This paper proposes a taxonomy laying out three possible approaches for thinking about risk thresholds in CHIM: 1) consistency; 2) engaging community perspectives; and 3) no risk threshold. We explore the strengths and weaknesses of each approach and offer a resource for those in the research community to inform their evaluation of ethical permissibility of current and future CHIM.