Results showed that half (50.5 percent) of the women in the cohort experienced an episode of recurrent IPV during the follow-up period. Among the 162 women who experienced recurrent IPV, 8 (5 percent) were threatened; 34 (25 percent) were slapped or pushed or had no injury; 33 (20 percent) were punched or kicked or had slight injury from some other form of abuse; 51 (31 percent) were beaten up, choked or burned or suffered serious injury from some other form of abuse; 28 (17 percent) were threatened with a weapon or experienced a severe injury from some other form of abuse; and 8 (5 percent) were seriously injured by a weapon. Findings identified a plausible set of predictors of the presence and severity of recurrent IPV from a cohort of low-income, inner-city women. The most provocative and potentially important finding is that leaving an abusive partner or successfully asking him to leave reduced the risk of recurrent IPV without increasing the risk of severe IPV in the multivariate model. However, since there was a moderate bivariate association between leaving an abusive partner and severity, the findings on leaving must be interpreted cautiously. Future longitudinal research should assess whether the findings from this study would be replicable in other samples of low-income women and generalizable to samples of non-indigent women. Data were collected from an inception cohort of 321 previously abused women from the Chicago Women’s Health Risk Study. Tables, references