Objectives The gender earnings ratio for year‐round full‐time (YRFT) workers varies substantially across U.S. states, with a range of 24 percentage points. I examine the sources of this variation to assess to what extent it reflects compositional differences by gender that vary across state and/or nonneutral effects of state of residence on gender earnings. Methods Using CPS data, I estimate earnings models for men and women that incorporate state fixed effects in addition to standard human capital and demographic variables. I use those estimates to compute unadjusted and regression‐adjusted estimates of the impact of state residence on the gender earnings ratio. Results I find that nonneutral gender‐specific state effects on earnings exist even after controlling for other determinants of earnings and that state of residence appears, therefore, to have a genuine effect on the gender earnings ratio. I also find that states with particularly low overall gender earnings ratios have consistently low ratios even within quite detailed education and occupation categories. Conclusions Variation in the gender earnings ratio for YRFT workers across states is not simply a result of compositional differences. It is unclear, however, what policy instruments or other factors account for these differences.