Abstract: The ability of making termination decision is an important index that influences the long pert formance of the venture capital firms. However, the present termination decision methods ignore the incoming information's effect during the venture project development process, thus influence the termination decision timd Aiming at this problem, from the view of Bayesian posterior estimate, this paper proposes the posterior probability signal learning process of venture capitalist firstly; based on determining the optimal investment amount to venture project, it then presents a model of venture project's optimal termination time based on venture project's value using the dynamic programming method; and a numerical example which analyses the optimal termination time model is given lastly.The optimal termination model put forward in this paper considers the influence of the incoming information releasing on the termination time, and it deals with the termination decision problem in the case of stages. It also provides reference for ventture capitalist to make termination decision. At the same time, this model can be also used to analyse ventture capitalist's optimal investment decision to venture project, it also provides reference for venture capeitalist's optimal decision. Besides, the analysis of model gives a reasonable explanation to the enterpriser's adverse selection behavior.

Key words: venture capital, optimal termination time, dynamic programming, the efficient boundary point