The invention discloses an upper limb rehabilitation training method based on a brain-computer interface and a virtual reality technology. A patient wears an electrode cap and a pair of VR glasses, and a computer, an EEG amplifier and a smartphone are connected; the computer judges whether the patient is trained for the first time or not; if yes, EEG signals are collected, individualized classifier calibration is performed, and then one training is started; otherwise, one training is directly started; in the training process, the pair of VR glasses and the smartphone construct an upper limb training action scene for a first-person perspective for the patient, the patient controls motions of the upper limb in a virtual scene in real time through continuous exercise imagination practice, an BCI module in the computer automatically adjusts the classifier according to the current training effect of the patient; after training is finished, the classifier of the BCI module performs self-adaptation adjustment for next training, it is ensured that training difficulty adapts to the patient functional recovery situation, data collection does not need to be performed before each rehabilitation training for calibration, and time is shortened.