Recent studies support the validity of both structured risk assessment tools and victim perceptions as predictors of risk for repeat intimate partner violence (IPV). Combining structured risk assessments and victim risk assessments leads to better predictions of repeat violence than either alone, suggesting that the two forms of assessment provide unique and complementary information. However, very little is known about elements involved in women's risk assessments. The present study explores predictors of women's risk assessment and differences in factors linked to victim and actuarial risk assessments in a large sample of women (N = 728) shortly after the arrest of their male partner for IPV. In multivariate analyses, women's risk assessments were strongly related to past relationship violence and their partner's substance abuse but weakly related to demographic factors, family constellation, and the partner's criminal history. Women who perceived high risk but had a low risk score on an actuarial measure were more likely to report the presence of dynamic risk factors, such as escalating violence and violence during separations, along with a history of emotional and psychological abuse. Qualitative findings paralleled quantitative findings, with women's stated reasons for expecting high or low risk indicating that women were attending to IPV history and dynamic factors. Implications for risk assessment and safety planning are discussed. (Contains 4 tables.)