Abstract: Wireless sensor network nodes generally have the characteristics of low cost, low power consumption, low rate, limited energy and so on. It is very sensitive to the changes of wireless channel to use inexpensive and low-power radio frequency (RF) modules for communication between sensors. In order to avoid wasted resources due to possible retransmissions on low-signal quality data links, methods such as routing protocols and redundant coding mechanisms are needed to continually evaluate the link quality of the network. For wireless sensor networks, both the adaptive redundancy coding and the routing protocol require a quantitative and accurate evaluation of the packet success rate (PSR). To solve this problem, we propose a method of received signal strength indicator (RSSI) based on the physical parameter to estimate the packet success rate of the wireless links. In the proposed method, the relationship between RSSI and PSR is obtained based on a large number of actual tests. Based on the RSSI value received in real time, the next-time PSR is estimated by Ornstein-Uhlenbeck model and Logistic curve model. Experiments and tests show that the proposed method has higher accuracy and fastness than the existing methods.