This study examined the gender-related differential predictive validity of five subscales of the Institutional Integration Scale (IIS) with regard to college student withdrawal. Differential functioning of the IIS across genders was assessed using an item response theory (IRT)-based framework of differential item and test functioning. The results confirmed the absence of differential functioning and supported the predictive validity of two of the five subscales for student withdrawal. IRT analyses revealed that a number of the items did not adequately reflect the construct and should be revised or removed from the measure. A discussion of these results and the implications for higher education institutions focused on preventing student withdrawal are presented. (Contains 5 tables.)