We analyze how quits responded to arbitrary differences in own and peer wages using an unusual feature of a pay raise at a large U.S. retailer. The firm's use of discrete pay steps created discontinuities in raises, where workers earning within 1 cent of each other received new wages that differed by 10 cents. First, we estimate a regression discontinuity (RD) model based on own wages; we find large causal effects of wages on quits, with quit elasticities less than -10. Next, we address whether the overall quit response reflects the impact of comparisons to market wages or to the wages of in-store peers. Here we use a multi-dimensional RD design that includes both a sharp RD in the own wage and a fuzzy RD in the average peer wage. We find that the large quit response mostly reflects relative-pay concerns and not market comparisons. After accounting for peer effects, quits do not appear to be very sensitive to wages – consistent with the presence of significant search frictions. Finally, we find that the relative-pay effect is nonlinear and driven mainly by workers who are paid less than their peers – suggesting concerns about fairness or disadvantageous inequity.