Two dependent variables were employed, the use of a gun and the use of a weapon other than a gun. In addition, correlates of self-defense weapon use by males and females and by whites and nonwhites were assessed. Results of logistic regression analysis suggested that gender and the presence of an offender's weapon were related to self-defensive weapon use. Additional significant predictors were distance from home and victim income for gun use only and number of recent moves, whether the attack occurred in the dark or during daylight, and time of day the incident took place for other weapon use only. Predictors for males and females and for whites and nonwhites varied, suggesting the need for continued exploration of separate models for these specific subgroups. Findings from NCS data confirmed the importance of incorporating elements of the total victimization context in research on victim responses and the need to include gender and race in the analysis of self-protective behavior. Data on univariate distributions of the survey data are appended. 34 references, 9 notes, and 4 tables