Abstract: With the dynamical features of mobility, intermittent connectivity and dynamic topology in delay tolerant network (DTN), the static random network coding methods become difficult to adapt to the applications in DTN environment. For working out this problem, a dynamic random network coding transfer method (DRNC) was proposed. Based on the Markov channel model, it first computed the transmit speeds of every network channel according to the receive data states of nodes, and then constructed the network flow graphic with channel capacities at the source node, calculated and forecasted the current network multicast capacity, after that, the random network coding scenario was extended and tailored according to the multicast capacity to transmit data packages under DTN environment. The simulation results show that, compared with the traditional random network coding method with fixed multicast rate, the dynamic random network coding method has reduced the average latency of data packages and increased the delivery probability.