Three studies examined if teachers' beliefs about writing predicted their efficacy to teach writing. We surveyed primary grade teachers from Taiwan (N = 782), Shanghai (N = 429), and the United States (N = 214). At each location, teachers completed surveys assessing attitudes toward writing and the teaching of writing, beliefs about students' progress as writers, and epistemological beliefs about writing instruction, writing development, and writing knowledge. We examined if each of these beliefs made unique and statistically significant contributions to predicting efficacy to teach writing after variance due to all other predictors, as well as personal and contextual variables, was controlled. With one exception, these three sets of beliefs each accounted for unique variance in predicting teacher efficacy at each location. There was, however, variability in unique variance in teacher efficacy scores accounted for by specific beliefs across locations and the factor structure of various measures by location.