Abstract. We prove global limit theorems for U -statistics and von Mises functionals. The rate of convergence in these theorems is also obtained.Limit theorems for U -statistics are usually considered for the uniform metric (see [1]-[3]). This paper is devoted to limit theorems for the mean metric for which we study the convergence of distributions of U -statistics and von Mises functionals.We prove global limit theorems and obtain the rate of convergence in these theorems for both degenerate and nondegenerate kernels.The idea of the proofs of the results is based on the method of metric distances (see [4]). There are other approaches used in the theory of U -statistics, such as the martingale method, the method of characteristic functions, the method of orthogonal decompositions, and some others. It is worthwhile to mention that V. M. Zolotarev ([4, p. 404]) was the first to point out that the method of metric distances works in the case of nonlinear models, too.The results of the paper are announced in [5].