A robust body of research has identified associations between positive teacher-child relationships--characterized by high levels of closeness and low levels of conflict--and children's academic achievement in elementary school (e.g. Roorda, 2012). Additional studies find that high-quality teacher-child relationships may promote academic resilience among lower-income, racial/ethnic minority children at-risk for poor achievement (Crosnoe et al., 2010; Murray & Zvoch, 2011). This work suggests that interventions designed to boost academic achievement in lower-income urban schools should consider targeting teacher-child relationship quality. Research, however, has yet to use multi-level models to infer causal impacts of high-quality teacher-child relationships on academic achievement within this high-risk population of students and schools. The current study uses a multi-level propensity score matching approach to estimate causal effects of a high-quality teacher-child relationship in kindergarten on student math and reading achievement at the transition to first grade. In this study, the researchers compare effects from these models to multi-level regressions, and conduct sensitivity analyses of continuous effects using generalized propensity score matching procedures. The results revealed sizeable, positive impacts of high-quality teacher-child relationships in kindergarten on a standardized measure of math achievement in first grade for a low-income, racial/ethnic minority population of students attending urban schools. However, no effects of high-quality teacher-child relations were detected for reading outcomes in first grade. Two tables are appended.