A problem of stabilisation of networked control systems (NCSs) with time-varying delay governed by homogeneous Markov chains, disorder of data packet and packet loss is discussed in this study. The time delay is decomposed into two parts: one is a fixed part which is an integer multiple of sampling period; the other is time-varying delay with more than one sampling period but bounded and governed by homogeneous Markov chains. Static delay-dependent controller which satisfies mean square stability is successfully designed using linear matrix inequalities techniques. Because controller designed based on a fixed transition probability matrix may not stabilised NCSs when time-varying delay changes along with networked load, the authors firstly design m controllers which satisfies different m transition probability matrices, and then constitute a stochastic switching controller with them and a switch. The switching rules of this stochastic switching controller are found through an intelligent algorithm based on greedy algorithm. Simulation example shown, if the switching controller would be used to stabilise NCSs, this system not only has larger stabilising span but also has better dynamic performances compared those has only one controller.