Results indicated that the structure of the MCVSI was best described by a hierarchical model that underpinned the superordinate general violence factors with two correlated but distinct forms of violent acts. It was also found that the questions on the MCVSI were not ordered correctly in terms of indicating severity of the underlying trait and that variation was evident in the discriminative power of different violent acts. Finally, the MCVSI was unable to accurately measure more serious acts of violence. The authors recommend that the MCVSI needs new items to improve the information about more serious acts of violence and they advise that violence measurement may be improved by disaggregating items into their component parts. The study involved administering the MCVSI via interview to 250 adult male prisoners housed in Scotland’s largest prison. The analysis focused on the frequency of violence and the researchers used a top-down and bottom-up approach to explore the underlying dimensional nature of the MCVSI. Four main problems evident in the research on violence measurement were analyzed: (1) violent acts were assumed to be underpinned by a unidimensional construct; (2) in terms of severity, the ordering of violent acts within a measurement instrument was frequently not based on empirical research; (3) different violent acts were treated as equally efficient indicators of violence; and (4) violence instruments might provide an accurate measurement for certain levels of violence but not others. Data analysis techniques included confirmatory factor analysis and nested chi-square tests. Follow-up studies should attempt to replicate these findings with different sample populations. Tables, figures, appendix, notes, references