While the practice of collecting information from applicants' professional references is widespread, there is a paucity of research linking references' assessments of applicants to subsequent performance. In this paper, we examine the predictive validity of a specific type of reference-provided information: categorical ratings of teacher applicants collected from their professional references -- a potentially low cost means of enhancing the applicant information available during the hiring process. We find an overall significant relationship between reference ratings and teacher performance as measured by observational evaluation ratings and teacher value-added in math, but that this relationship is moderated by two factors. First, while references' ratings of applicants with prior teaching experience are predictive of performance, those of novice applicants are not. Second, the predictive validity of reference ratings varies according to rater type: ratings from references identified as the applicants' "Principal/Other Supervisor," "Instructional Coach/Department Chair," or "Colleague" are significantly predictive of performance while those from other types of raters are not. Overall, our findings show that meaningful information can be solicited from applicants' references in the form of categorical ratings but also demonstrate some limitations in the potential for this type of information to inform hiring decisions. [This is the online version of an article published in "Education Finance and Policy."]