Previous research suggests that anger has important social and health consequences, particularly cardiovascular health. The pathogenic aspects of anger have not been identified, however, in part because of a reliance on unidimensional measures of anger. The present article describes psychometric data for an inventory that is sensitive to the multidimensional nature of the anger construct. It was hypothesized that the newly developed Multidimensional Anger Inventory (MAI) would include scales reflective of the following dimensions of anger: frequency, duration, magnitude, mode of expression, hostile outlook, and range of anger-eliciting situations. The mode of expression dimension was expected to contain separate anger-in, anger-out, guilt, brood, and anger-discuss measures. The inventory was administered to two populations: male and female college students and male factory workers. Factor analyses of the MAI within the two samples showed that the frequency, duration, and magnitude dimensions clustered together to form an anger-arousal factor that accounted for 64% and 71% of the variance in the two samples, respectively. The range of anger-eliciting situations and hostile outlook emerged as separate dimensions, as hypothesized. Mode of anger expression was best described by two dimensions labeled anger-in and anger-out. Psychometric analyses of the scale showed that it possessed adequate test-retest reliability (r = 0.75) and high internal consistency (alpha = .84 and .89 for the two samples). The validity of the scale was supported by the expected pattern of relations with other inventories designed to assess anger or hostility. Comparisons of MAI scores between (college versus factory) and within (male versus female) populations were made.