Development and initial validation of a scale for assessment of agitation in traumatic brain-injured individuals is described. Items were selected from an initial 39-item pool based on their testability, differentiation of agitation, frequency of occurrence, and representation of the full domain of the construct. The resulting 14-item instrument, called the Agitated Behavior Scale (ABS), was subsequently tested on an independent sample of 35 head-injured subjects. Reliability was examined via measures of internal consistency and qualitative evaluation of principal components factor analysis. Cronbach's alpha exceeded .80 for all rates. Comparable values of theta were found and qualitative criteria of internal consistency were met. The ABS score accounted for between 36% and 62% of the variance in 15 of 16 correlations with simultaneous, independent observations of agitation. Results of initial validation are discussed in light of classic test theory and research necessary for determination of construct validity.