Incivility is becoming highly relevant in online political communication on social media, and existing scholarship has emphasized the importance of identifying uncivil content and its downstream consequences. At the same time, scholars agree that incivility is a challenging concept to define as, among other things, it is context-dependent. As a result, operationalizations vary widely in the literature, ranging from measurements of the concept as a continuous scale to its decomposition into different dimensions and its decoupling from concepts like intolerance. While this research is growing, we still have insufficient understanding as to whether, for example, intolerance, incivility, and other forms of toxic and threatening behaviour are separate dimensions of undesirable behavior. While past work has shown that incivility and intolerance can occur in different contexts, when it comes to the consequences for individuals, is intolerance an analytically distinct concept from incivility and when exactly do citizens perceive comments as “crossing a line”? We propose a new theoretical framework covering the dimensions incivility, intolerance, and threat to explain the effects of incivility on emotional state, activation, and preferences for moderation. Using an original online experiment, we attempt to uncover the consequences of uncivil online discussions. We build on recent theoretical developments on incivility and expose participants to political social media posts varying in the level of incivility (civil versus uncivil), tolerance (tolerance versus intolerance) and threat (high threat versus low threat). Our findings contribute to the current debate with a more nuanced definition of incivility and add important measurement considerations.