The problem of state estimation for neural networks with two different Markovian jumping parameters is studied. The Markovian jumping parameters in connection weight matrices and discrete time-varying delay are assumed to be different. Leakage delay is added to the considered model. Matrix decomposition method and Wirtinger-based integral inequality are introduced to obtain a less conservative condition. An appropriate Lyapunov–Krasovskii function is constructed which contains more mode-dependent Lyapunov matrices.