Physics, as a foundational science, has particular importance in predicting the postsecondary success of students who major in science, technology, engineering, and mathematics. This quasiexperimental, observational study examined teacher-level and school-level predictors of student performance in physics, with a focus on isolated teachers. A New York State case study is useful since the teacher certification policy is largely determined at the state level in the U.S. The overall sample included New York State public schools that offered physics (N ¼ 960), physics teachers (N ¼ 1584), and student physics test takers (N ¼ 47 734) in the academic year 2016-2017. Teacher-level variables included the content preparation and certification of physics teachers, physics course load, professional age (years of experience), whether the teacher was isolated, whether the teacher taught mathematics, and whether the teacher taught Advanced Placement Physics; and school-level variables including physics standardized test passing rates, school size, socioeconomic status, locale, and physics course taking ratio. Data were collected from a variety of publicly available sources that were verified by state education agencies. Results indicated a significant proportion (40%) of physics teachers were isolated, and their students tended to have weaker physics performance scores than students of nonisolated teachers. Compared to the nonisolated teachers, a larger percentage of isolated physics teachers were uncertified in physics and taught in urban and rural schools. There was no significant difference in professional age between isolated and nonisolated teachers, but urban teachers had less teaching experience than suburban and rural physics teachers. When analyzing the subset of isolated teachers (n ¼ 449), a multiple linear regression model indicated urban locale and schoollevel socioeconomic status were the main negative predictors of student physics performance, while rural physics locale and professional age were positive predictors of physics performance; the model explained 38% of the variance, a large effect. Teaching experience acted as a mediator of poverty and urban locale in predicting student physics performance with a small effect size. Implications related to equity considerations and physics education policy are discussed.