The subjects were 73 adult male volunteers in a minimum security prison. On the basis of their current offense, 22 percent were classified as drug offenders, 59 percent as property offenders, 4 percent as nonviolent sex offenders, and 15 percent as violent offenders. The percentages were similar to those for the prison's entire population. The subjects completed a video taped measure of preferred IPD (Walkey and Gilmour, 1979), and prison records were then searched for information leading to the classification of each subject as either violent or nonviolent and for data on 17 background variables. A discriminant function analysis identified the following variables, in the order of their inclusion, as discriminators between violent and nonviolent offenders: current offense, group offense, IPD, family background, height, previous convictions, and alcohol related to offense. An inspection of discriminant scores revealed an optimal criterion score with which 92 percent of the subjects were correctly identified as either violent or nonviolent. A second analysis, which omitted the variable, 'current offense,' showed that IPD was the next best discriminator between violent and nonviolent offenders, and a positive relationship between preferred IPD and violence was clearly demonstrated. The results also supported studies that have shown no difference in the shapes of the body-buffer zones of violent and nonviolent offenders, but anticipated differences between the IPD scores of Polynesian and European ethnic groups were not found. Further studies which would establish whether offenders with a larger preferred IPD are more likely to engage in violent acts in prison than those with lower IPD scores are suggested. Related studies are reviewed. Data tables and a 21-item reference list are included. (Author abstract modified)