Abstract This paper addresses the issue of mean-square asymptotic synchronization (MSAS) of complex dynamical networks with communication delay and switching topology. The communication delay is assumed to be time-variant and bounded, and the switching topology is governed by a semi-Markovian process and allowed to be asymmetric. A distributed control law based on state feedback is presented. Two criteria for the MSAS are derived using a mode-dependent Lyapunov-Krasovskii functional, the Bessel-Legendre integral inequality, and a parameter-dependent convex combination inequality, for the asymmetric and symmetric topology cases, respectively. The scenario of fixed topology is also considered, for which two asymptotic synchronization criteria are proposed. Two simulation examples are provided to illustrate the effectiveness and reduced conservatism of the proposed theoretical results.