In the infrared small target detection system, CFAR (Constant False Alarm Rate) is a commonly used technology, but in the traditional single frame detection method, detection rate is requested to be improved while the false alarm rate is increasing. This paper proposes a threshold attenuation CFAR detection method based on Gauss distribution. After the preprocessing of infrared images, we came into the designing of CFAR detector based on Gauss distribution. According to the previous frame target location and attenuation of local threshold, the detection rate of the target neighbourhood can be improved to obtain the current target location. The experimental results show that the proposed method can effectively control the threshold, and under the precondition that the background clutter was suppressed by the global low false alarm rate, it can improve the local detection rate and reduce the probability of target loss.