For a broad class of symmetric weakly interacting particle systems, it has been shown that as the number of particles goes to infinity, the empirical measure process converges to the law of a nonlinear Markov process. Similar results have been established for the empirical measure of the Nash equilibria of symmetrically interacting agents, where each agent behaves strategically, in which case the convergence is to the solution of a corresponding mean-field game. We establish non-asymptotic concentration bounds that, in particular, provide convergence rates in these limit theorems. We also describe large deviation results for mean-field games, both in the presence and absence of common noise. This talk is based on joint work with Dan Lacker and Francois Delarue.