The invention discloses an infrared temperature measurement calibration method. The method is characterized by through introducing an environment temperature variation and a temperature measurement distance, compensating a temperature measurement deviation brought by a temperature drift, in a temperature-regulated box, continuously changing an environment temperature in the box and using a black body to carry out temperature calibration, and acquiring a first corresponding relation between an image gray scale capable of reflecting an infrared heat radiation energy size and the environment temperature; in addition, correspondingly acquiring a second corresponding relation between the image gray scale and a black body temperature and a third corresponding relation between the image gray scale and a temperature measurement distance; and finally, fitting the first corresponding relation, the second corresponding relation and the third corresponding relation through a BP neural network andcalculating temperature information of an infrared temperature measurement system heat radiation target through fitting. A problem that temperature measurement precision is low because of the temperature drift caused by an environment temperature change is overcome, infrared temperature measurement precision is increased, and a false alarm rate of fault detection is reduced.