Examined the ability of demographic variables and risk factors (parental su
bstance use, number of people in the home, out-of-home placements, grades r
epeated, arrest history, and total life stress) to predict exposure to comm
unity violence among 342 high school students from inner-city Baltimore ref
erred for mental health care in community centers or in the schools. Over 9
0% of the sample knew at least 1 victim of a violent act, 77% reported witn
essing a violent act, and 47% reported past victimization by violence. Risk
variables were more powerful regression predictors of violence exposure th
an demographic characteristics such as race, sex, or clinical setting. Even
after controlling for demographic differences in violence exposure, risk f
actors as a group accounted for another 10% to 15% of variance. Life stress
was the most consistent predictor of violence exposure for this sample, an
d life stress was the only variable to make a significant unique contributi
on to the prediction of all 4 violence criteria.