Irritability, characterized by a tendency to exhibit increased anger, is a common clinical problem in youth. Irritability is a significant clinical issue in youth with various psychiatric diagnoses, especially disruptive behavior, and mood disorders (Attention-Deficit/Hyperactivity Disorder, Oppositional Defiant Disorder, Conduct Disorder, and Disruptive Mood Dysregulation Disorder). Although there have been previous studies focusing on functional alteration in the amygdala related to irritability, there is no comprehensive model between emotional, neuronal, and behavioral characteristics. Using an functional magnetic resonance imaging (fMRI) procedure, we investigated the relationships between behavioral irritability, selective impairments in processing facial emotions and the amygdala neural response in youth with increased irritability. Fifty-nine youth with disruptive mood and behavior disorder completed a facial expression processing task with an event-related fMRI paradigm. The severity of irritability was evaluated using the Affective Reactivity Index. In the result of behavioral data, irritability, and reaction time (RT) differences between interpreting negative (fear) and positive (happiness) facial expressions were positively correlated. In the fMRI result, youth showed higher activation in the right cingulate gyrus, bilateral cerebellum, right amygdala, right precuneus, right superior frontal gyrus, right middle occipital gyrus, and middle temporal gyrus, during the happiness condition vs. fear condition. No brain region exhibited greater activation in the fear than in the happiness conditions. In the result of the mediator analysis, increased irritability was associated with a longer RT toward positive vs. negative facial expressions. Irritability was also positively associated with the difference in amygdala blood oxygen level-dependent responses between the two emotional conditions (happiness > fear). This difference in amygdala activity mediated the interaction between irritability and the RT difference between negative and positive facial expressions. We suggest that impairment in the implicit processing of facial emotional expressions with different valences causes distinct patterns of amygdala response, which correlate with the level of irritability. These results broaden our understanding of the biological mechanism of irritability at the neural level and provide information for the future direction of the study.