How much should judges be paid? We first survey the considerable history of the debate and identify the implicit causal claims made about the effect of judicial pay. We find that claims about the effect of pay on the composition and quality of the judiciary have remained remarkably similar over the past two hundred years. In contrast, claims about the effect of pay on judicial independence have changed as the meaning of judicial independence itself has shifted. We take advantage of the large variation in real salaries and opportunity costs for state appellate court judges across states from 1977 to 2007 to empirically test these claims. We find that judicial salaries have a small but significant effect on the likelihood of exit and thus the length of judicial tenure, and a small effect on the background of judges that join the appellate bench. A more limited analysis of California trial court judges finds far more sensitivity to pay, however, suggesting that trial and appellate court judges may behave differently.