Giant panda population data are important bases for knowing the population dynamics of giant pandas. Thus, it is significant to model giant panda population data for conservation. To solve the problems that the existing methods for modeling population dynamics were not able to capture the randomness in a complex ecological system and have bad extensibility, a probability membrane system for modeling the giant panda ecosystem is proposed based on the data from Chengdu Research Base of Giant Panda Breeding. On the basis of giant panda pedigree, a probability membrane system is designed with a membrane structure consisting of two nested membranes, a series of objects and evolution rules for representing the giant panda ecosystem. The simulation results show that the model can reflect the dynamic changes of giant panda population and therefore can provide reference for managers.