The relative motion between the moving parts and the shell of the checkweigher and the resonance of mechanical natural frequency point under external force result in the vibration disturbance. To address this problem, the variable step size least mean square (LMS) adaptive notch filter based on improved versiera algorithm is proposed. Compared with the traditional antivibration methods, such as movingaverage filter, Butterworth lowpass filter, Butterworth notch filter and fixed step size LMS notch filter, the variable step size LMS adaptive notch filter is proved to be accurate and superior in filtering vibration interference by analyzing the checkweigher′s vibration characteristics. The method combined with movingaverage and asymmetric tailcutting mean filtering is used to finalize the result on the checkweigher at various speeds. Experimental results show that the average error is ≤0171 g and the standard deviation is ≤0240%, which can meet the requirement of category XIII(1).