Purpose of the Study: Ambivalence has been described as simultaneous positive and negative emotional experiences. Although ambivalent feelings are often reported by dementia family caregivers, the effect of these feelings on caregivers' mental health has not been studied. Furthermore, the measurement of ambivalence specific to caregiving situations has not been studied. The aims of this study are to analyze the psychometric properties of the Caregiving Ambivalence Scale (CAS) and, drawing upon the stress and coping model, to test whether ambivalent feelings significantly contribute to caregivers' distress. Design and Methods: Participants were 401 dementia family caregivers. Face-to-face interviews were conducted which included measures of ambivalence, depressive (CES-D) and anxious symptomatology (POMS), stressors (disruptive behaviors subscale of the RMBPC), and sociodemographic information. Results: Results from exploratory, parallel, and confirmatory factor analyses suggest that the CAS has a unidimensional structure, explaining a 64.26% of the variance of ambivalent feelings. Good reliability and validity indexes were found, including a Cronbach's alpha of .86. The results revealed significant (p < .01) positive associations with depressive and anxious symptomatology, and frequency and reaction to disruptive behaviors. Ambivalent feeling scores significantly contributed to the explanation of caregivers' depressive and anxious symptoms after controlling for sociodemographic and stressor variables. Implications: The CAS shows good psychometric properties that recommend its use as a measure of ambivalent feelings in caregivers and appears to be a relevant variable for understanding caregivers' mental health.