Brief, gentle arm restraint is widely used in experimental studies of children's anger, but the pattern of responses generated by such restraint has been incompletely described. We now describe a hierarchy of responses within trials as well as an escalation across trials that have both methodological and theoretical significance. Mothers of 87 15-month olds prevented them from playing with a toy by restraining their arms on two consecutive 30 sec trials. Physical struggling was the first and most frequent response; children who struggled were significantly more likely to vocalize, and those who vocalized were significantly more likely to show facial expressions of anger. The children's responses became more probable, rapid, and intense during Trial 2 restraint. Overall, the hierarchy was orderly enough to meet criteria for Guttman scalability. The particular sequence observed suggests situational, as opposed to bio-energetic, ordering of responses. Methodologically, the two trial paradigm is a simple, ecologically valid model for studying anger escalation that parallels the "attack priming" of aggression in other species. The magnitude and persistence of anger priming may provide novel measures of anger regulation. Theoretically, the existence of an orderly response hierarchy is consistent with previous observations suggesting that, within a situational context, the sequential appearance of specific behaviors may indicate progressive increases in anger intensity.