A human body flexibility detection and immersion type virtual training method of the invention can solve the technical problems that at present, flexibility level detection and training modes are notscientific, interestingness is low, and digital feedback does not exist. The method comprises the following steps that S100, an acquisition terminal firstly acquires and tests the flexibility of a human body, a series of actions are set in the testing process to detect the level of the flexibility of the human body, and a training scheme is generated on the basis; S200, a training scheme adopts acool form, different obstacles are set on a path, and the action requirements of the obstacles are determined according to the level of the human body test in the step S100. In the using process, interest of people in flexibility level training can be greatly improved, and compared with optical posture capturing such as motion sensing games, the method has higher stability, is not affected by theenvironment, crowds, body types and clothes and has better applicability.