This talk will introduce the central concepts of my account of toxic speech, drawing from both philosophy of language and social epidemiology. A medical conception of toxicity looks at the toxin, the dose, frequency, route of delivery, susceptibility of the individual or population, and more. All these factors can be fruitfully explored in connection to understanding the mechanisms by which speech acts and discursive practices can inflict harm, making sense of claims about harms arising from speech devoid of slurs, epithets, or a narrower class I call ‘deeply derogatory terms.’ By highlighting the role of uptake and susceptibility, and this model suggests a framework for thinking about damage variation. Toxic effects vary depending on one’s epistemic position, access, and authority. An inferentialist account of discursive practice plus a dynamic view of the power of language games offers tools to analyze the toxic power of speech acts. Even a simple account of language games helps track changes in our discursive practices. Identifying patterns contributes to an epidemiology of toxic speech, which might include tracking increasing use of derogatory terms, us/them dichotomization, terms of isolation, new essentialisms, and more. Finally, I will highlight a few ways to respond to toxic speech, some predictable from past theories (Brandom’s challenging, Langton’s blocking) and some perhaps more unexpected.