Objective: To examine child facial emotion recognition accuracy (ERA) in high-risk for child physical abuse (CPA) parents and low-risk for CPA parents (Study 1) and to conduct a meta-analysis summarizing published research on the relationship between child facial ERA and CPA (Study 2). Method: In Study 1, ERA data for child facial emotions were obtained from mothers and fathers who were at high risk (n = 51) or low risk (n = 61) for CPA. In 1 of 2 presentation time conditions (100 ms, 600 ms), parents evaluated child photographs, taken from the Radboud Faces Database, which displayed 5 face emotions (angry, happy, sad, fearful, neutral) at 3 face angles (frontal, 45 degrees, 90 degrees). In Study 2, a meta-analysis of published studies was used to estimate the overall effect size of ERA differences between high-risk/abusive and comparison parents. Results: In Study 1, ERA differences were found for emotions (largest ERA for happiness), face angles (frontal > 45 degrees > 90 degrees), and presentation times (100 ms < 600 ms); however, only an overall trend for ERA risk group differences was observed. Nevertheless, the Study 2 meta-analyses revealed a significant effect size reflecting an overall moderate ERA difference between high-risk/abusive and comparison parents, and the effect size was not moderated by the population studied (high-risk parents vs. abusive parents). Conclusions: Because child facial ERA appears to be associated with CPA, the manner in which parental child emotion recognition errors contribute to problematic parent–child interactions merits additional study. (PsycINFO Database Record (c) 2015 APA, all rights reserved)