Why do productive workers and firms locate together in dense cities? I develop a new theory of two-sided sorting in which both heterogeneous workers and firms sort across space. The location choices of workers and firms affect each other and endogenously generate spatial disparities in the presence of three essential forces: complementarity between worker and firm productivity, random matching within frictional local labor markets, and congestion costs. I demonstrate that the decentralized equilibrium exhibits excessive concentration of workers and firms, and dispersing them away from dense locations can mitigate congestion without reducing output. I then provide direct empirical evidence of the two-sided sorting mechanism using German administrative microdata. An exogenous increase in the quality of the workforce in a location results in more productive firms choosing that location. Finally, to quantify the implications of the model, I calibrate it to U.S. regional data and show that policies that relocate workers and firms toward less dense areas can increase welfare.