We estimate the impact of nurse unions on health care quality using patient discharge data and the universe of hospital unionizations in California between 1996 and 2005. We find that hospitals with a successful union election outperform hospitals with a failed election in 12 of 13 nurse sensitive patient outcomes measures. We also find that hospitals with a unionization drive are establishments with declining quality as measured by patient outcomes. When such declines are accounted for using hospital-specific trends, we find that unionized hospitals also outperform hospitals without any union election in the same 12 of 13 outcome measures. The timing of the quality improvement is consistent with a causal impact: the largest changes occur precisely in the year of unionization. The biggest improvements are found in the incidence of metabolic derangement, pulmonary failure, and central nervous system disorders such as depression and delusion, where the estimated changes are between 15% and 60% of the mean incidence for those measures. Dynamic estimates confirm that the improvements in health care outcomes occur within the first two years following nurse unionization.