Most SSVEP (Steady-State Visual Evoked Potentials) based BCI (Brain Computer Interface) systems rely on computer displays, which are not convenient to integration so the application of visual-based BCI is limited in daily life. VR (Virtual Reality) technology not only can provide a realistic 3D (three-dimension) visual stimulus, but also has wearable and portable devices integrated with independent monitors, which has great potentials for the applications of visual-based BCI. Therefore, this paper introduces a VR glass to display 3D-SSVEP stimuli. By modelling a 3D-SSVEP based BCI, the subjects can be deeply immersed in BCI systems. To validate the performance enhancement of 3D-SSVEP based BCI, this paper used four classification methods to compare the target recognition results between 3 and 2D visual stimuli, and the experiment results showed that the highest recognition accuracies of 3D and 2D paradigms by CCA, MSI, FBCCA, and TRCA were 93.26% and 87.12%, 93.71% and 85.51%, 96.52% and 91.97%, and 98.72% and 96.3% respectively. Moreover, under the 3D visual stimuli with a duration of 1 s, the TRCA method achieved a maximum information transfer rate of 78.15 bits/min. The results indicated that 3D-SSVEP by VR performed better than 2D paradigm, and TRCA showed the best recognition performance. This research verified that 3D visual stimuli can effectively enhance the target recognition performance of SSVEP, which provides new insights for the application of VR devices in the visual-based BCI systems.