This study utilizes the binomial hierarchical generalized linear modeling (HGLM) technique and nationally representative data (SASS 2003-2004) to examine the relationship between teacher unions and teacher quality in America's large urban and suburban districts and the effect of teacher unions on the intra-district distribution of teacher quality across schools with varying poverty and minority student concentration in the largest districts. Results reveal that compared with non-unionized districts, strongly unionized districts tend to have higher proportions of NCLB defined highly qualified teachers, teachers with at least five years of experience, teachers with subject-area degrees, and teachers with subject-area certifications in the large urban and suburban districts. But, strongly unionized and non-unionized districts have comparable proportions of empirically-defined high quality teachers and teachers who graduated from selective colleges. Weakly unionized districts are less likely to attract and retain experienced teachers than non-unionized ones. This study also finds that in the largest districts school poverty/minority level has a stronger (and negative) effect on the distribution of experienced teachers in strongly unionized districts than in non-unionized districts, which suggests that in strongly unionized districts the teacher quality gap is much wider across high and low poverty/minority schools in terms of employing experienced teachers.