The research sample consists of 468 convictions resulting from 619 indictments for the crimes of murder, rape, aggravated assault, and robbery. Independent variables examined include seven offender and case-processing characteristics. The dependent variable, sentence severity, is measured first as a dichotomous variable (probation or prison) and then as an interval variable (prison-sentence length). After determining that crime type explains a significant proportion of the total variance in sentence severity, the dependent variables are than regressed on the variable list. Four regressions are performed for each dependent variable; the first analysis uses all cases in the sample, while the remaining analyses focus separately on the crimes of murder, assault, and robbery. The research indicates that the aggregate and crime-specific analyses produce substantively different results, both in terms of the factors found to influence sentence severity and the total amount of variance explained. Findings demonstrate the utility of crime-specific analysis. (Author abstract)