This paper proposes a new method of 3D indoor positioning based on Bluetooth devices. Virtual Received Signal Strength Indicator (RSSI) fingerprint is built on RSSI difference and distance ratio. The concepts and stability of 1D and 2D virtual fingerprints are introduced, and the property of uniqueness for virtual fingerprint is proved. 2D virtual fingerprint is more stable than that of 1D for indoor positioning. Various positioning algorithms, based on virtual fingerprint and one-way ranging, are discussed, such as NN, KNN, Multi-Step and nonlinear optimization, inference and regression. Multi-Step algorithm holds the best performance in time and error. The proposed method utilizes virtual fingerprint and two-way ranging to overcome the four evident shortcomings of the traditional RRSI fingerprint-based methods. Firstly, measured RSSI is calibration-less for different Bluetooth devices and varying battery status. Secondly, virtual fingerprint of training is obtained by computing, not measured point by point in whole 3D indoor environment. Thirdly, 3D indoor positioning method is employed to substitute for the classical 2D positioning methods. Finally, two-way ranging is adopted to take the place of the conventional one-way ranging and improve the positioning precision. The correctness and availability of the proposed method are testified by the simulation and hardware experiments.