Abstract: Additive manufacturing TC4 alloy is a kind of metal material with excellent mechanical properties and process properties. It has been widely used in the aerospace field. In recent years, the effect of stress state on the deformation and failure behavior of metal materials has been widely concerned in the research field of plastic mechanics. However, most of these studies were performed considering quasi-static state only, but few of them took into account the high strain rates. In this paper, the effects of stress state on the deformation and failure behavior of additive manufacturing TC4 alloy were systematically investigated. Starting from the basic mechanical properties of additive manufacturing TC4 alloy, the stress triaxiality \eta and Lode angle parameters \overline \theta are used to characterize the stress state. Using electronic universal testing machine, high-speed hydraulic servo testing machine and split Hopkinson bar, combined with digital image correlation analysis, the mechanical properties of additive manufacturing TC4 alloy under different strain rates and different stress states were tested. The deformation and failure characteristics of the material under various working conditions are obtained. In order to obtain the internal stress state history parameters and strain field of the sample, this article passes ABAQUS performs numerical simulation to obtain the stress state history parameters and failure strain at the maximum strain of the specimen. Based on the results of experimental testing and simulation analysis, the traditional MMC failure model was revised, and the failure model of material was established that fully considered the strain rate, stress triaxiality and Lode angle effect. At the same time, a Johnson-Cook failure model considering the effects of stress triaxiality η and strain rate is established. A high-speed impact experiment was carried out on the additive manufacturing TC4 alloy plate, and numerical simulation was carried out for the experiment, which verified the usability of the established constitutive model and failure model.