Objective Performance thresholds are commonly used in pay‐for‐performance (P4P) incentives, where providers receive a bonus payment for achieving a prespecified target threshold but may produce discontinuous incentives, with providers just below the threshold having the strongest incentive to improve and providers either far below or above the threshold having little incentive. We investigate the effect of performance thresholds on provider response in the setting of nursing home P4P. Data Sources The Minimum Data Set (MDS) and Online Survey, Certification, and Reporting (OSCAR) datasets. Study Setting and Design Difference‐in‐differences design to test for changes in nursing home performance in three states that implemented threshold‐based P4P (Colorado, Georgia, and Oklahoma) versus three comparator states (Arizona, Tennessee, and Arkansas) between 2006 and 2009. Principal Findings We find that those farthest below the threshold (i.e., the worst‐performing nursing homes) had the largest improvements under threshold‐based P4P while those farthest above the threshold worsened. This effect did not vary with the percentage of Medicaid residents in a nursing home. Conclusions Threshold‐based P4P may provide perverse incentives for nursing homes above the performance threshold, but we do not find evidence to support concerns about the effects of performance thresholds on low‐performing nursing homes.