We consider Lagrangian systems in the limit of infinitely many particles. It is shown that the corresponding discrete action functionals Γ-converge to a continuum action functional acting on probability measures of particle trajectories. Also, the convergence of stationary points of the action is established. Minimizers of the limiting functional and, more generally, limiting distributions of stationary points are investigated and shown to be concentrated on orbits of the Euler–Lagrange flow. We also consider time discretized systems. These results in particular provide a convergence analysis for optimal transportation meshfree methods for the approximation of particle flows by finite discrete Lagrangian dynamics.