Aggression and violence are significant clinical problems in psychiatric inpatient units. However, research exploring violent behavior is limited by the lack of an adequate research scale. This paper reports on the development and testing of the Violence Scale (VS), a behavioral rating scale which measures aggressive and violent behavior towards self, others and property. The VS has been tested in two studies of psychiatric inpatients (N = 162) and (N = 42). Tests of reliability included internal consistency, item analysis and stability. Tests of validity included content and construct validity. The psychometric results were evaluated through the application of standard statistical criteria [Carmines, E. G. and Zeller, R. A. (1979). Reliability and Validity Assessment. Sage University Press, Beverly Hills, CA; Cronbach, L. J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika 16, 197-333; Nunnally J. (1978). Psychometric Theory, McGraw-Hill, New York]. Coefficient alpha was 0.91 (study 1) and 0.68 (study 2) and stability was r = 0.79 (study 1). The items met most of the established criteria for item analysis and internal consistency. In factor analysis, the items met the criteria for loading onto the three predicted factors (others, self and property). Results of predictive model testing indicated that three of the four predicted relationships were supported. Initial testing indicates the scale is moderately stable, is internally consistent, and has evidence of initial limited construct validity.