Ultrasonic sensing systems are currently used in automobiles and mobile robots. Recently, under conditions of multiple ultrasonic sensing systems operating in a small area, we proposed an ultrasonic multiple access method (UMAM) based on the spectrum-spread technology using M-sequence code. However, the effect of the UMAM depends on the preset threshold level. In this paper, we consider an automatic interference elimination algorithm for UMAM. Under conditions with a known interference-signal ratio, the normalized correlation function between two microphones can be used to investigate the effect of the threshold level in UMAM, and further, the optimal threshold level can be obtained from the maximum value of the normalized correlation function. Using a set of suitable training examples, the approximation function between the optimal threshold and the interference-signal ratio can be obtained. From the experimental results, the automatic interference elimination algorithm is a useful tool.