We use US longitudinal survey data to examine the role of performance pay (other than profit sharing) in worker quit decisions. We argue that performance pay should increasingly be viewed as an indicator of an internal labor market rather than of a simple contemporaneous incentive. Suggestive of this claim, we find that in ever more complete specifications that account for worker and employer characteristics, aggregate earnings and worker job satisfaction, performance pay is associated with a reduced probability of worker quits. This remains when including worker fixed effects that control for unmeasured invariant heterogeneity. We investigate how it varies with the type of performance pay and its intensity. We confirm heterogeneity in this influence by workplace size.