The skill gap in geographical mobility is entirely driven by workers who report moving for a new job. A natural explanation lies in the large expected surplus accruing to skilled job matches. Just as large surpluses ease the frictions which impede job search in general, they also help overcome those frictions (specifically moving costs) which plague cross-city matching in particular. I reject the alternative hypothesis that mobility differences are driven by variation in the moving costs themselves, based on PSID evidence on self-reported willingness to move. Evidence on wage processes also supports my claims.