Based on information produced by a 200 kHz ultrasonic echo device developed for robotics and other industrial purposes, a method for object identification was developed and is presented in this paper. Interpretation of the echos from the reflecting objects, complexity analysis and information extraction were made by system identification methods such as impulse response analysis and state-space realization. A clustering-type discrimination of different objects based on the echo identification result was made in a subsequent step using the least-squares method. Experimental evaluation showed that the method is effective for object recognition.