<p>Most protests do not become violent, yet some protest crowds become increasingly confrontational as events unfold. Explanations based on deindividuation remain limited because they do not adequately explain why escalation is selective, target-specific, and temporally patterned. A more precise question concerns not why one group is violent and another is not, but how the same protest crowd can change over time. The present article develops a contextual appraisal account of within-crowd escalation. Drawing on the Interaction Discrepancy Model (IDM) and situating it alongside the Elaborated Social Identity Model (ESIM), protest escalation is conceptualized as a process of reappraisal. Crowds often begin with anger: an undesirable discrepancy is attributed to a blameworthy other who is still seen as pressureable, making retributive confrontation more likely. Escalation deepens when failed redress, dismissive treatment, or illegitimate force revise the crowd’s judgment of changeability. As the target comes to be seen as corrupt, incorrigible, or beyond repair, anger may harden into disgust, shifting action from coercive pressure toward expulsive or destructive tactics. Two secondary case studies are used to examine this sequence: the 1999 Seattle WTO protests and the 2011 Tottenham protest-turned-riot. Both cases support a dynamic, interactional account of escalation and fit poorly with a simple deindividuation model. At the same time, they indicate that anger and disgust should be treated as ideal–typical phases within a broader process shaped by identity, policing, organization, and political context.</p>