Abstract: An autonomous virtual hand behavior construction method for virtual manufacturing environments is presented to improve the current mainstream virtual hand animation method which totally depends on capturing a user’s hand movement. The goal of this research is to provide an intelligent efficient virtual hand for virtual manufacturing scenario. The perception ability of virtual hand is established by adding ring sensors to its structural model. An adaptive behavior mechanism of virtual hand is constructed to alleviate user’s cognitive load and manipulative loads which is composed by three-stage perception algorithms the authors invented. Furthermore, a gesture set and a gesture state transition model are given to realize a continuous grasping behavior of virtual hand on the surfaces of complex parts. The focus of this work is to investigate how to construct and realize a kind of continuous virtual hand manipulation behavior mechanism and algorithms on the surfaces of complex parts. The experiment results show that the designed virtual hand in this paper has a good geometric realism and behavior realism. The method given in this paper has characteristics of usability, reliability and real-time.