Although the least square method is widely used in the calibration of infrared thermal cameras at present, its accuracy is not ideal. To reduce the influences of the system random error, environmental fluctuation and unstable blackbody temperature of a thermal camera on the calibration result, a calibration model for infrared thermal cameras based on a weighted least square method is established. Both calibration experiments and radiation measurement experiments are carried out by the weighted least square method and the least square method respectively. The experimental results show that the calibration model based on the weighted least square has a higher fitting accuracy and its mean measurement accuracy in radiation measurement experiments is improved by 2.3% compared with that of the least square method. The calibration based on the weighted least square method is of practical value and of a certain referential significance.