A modest body of research has examined judicial decision making in civil protection order (CPO) cases. A major finding of this prior research is that the factors expected to shape judicial responses to CPO requests are often found to be insignificant. Because such decisions are often rendered in an environment of vast judicial discretion and competing allegations, the question of "what matters?" assumes added importance. This study examines permanent/final restraining order (PRO) outcomes for intimate partner violence on a number of variables. Specifically, chi-square analyses were performed examining the associations between granting/denying a PRO and demographic, relationship, hearing, and allegation characteristics associated with the petitioner and respondent. These tests helped to reveal relationships at the bivariate level and aided in further model-building using logistic regression and decision-tree analysis. The findings show that the factors most associated with PRO outcomes, namely, the denial of a PRO, are those reflecting the licit rather than illicit behavior of the respondent.