In this article nut-shell pieces are separated from pistachio kernels and hazelnut kernels using impact vibration analysis. Vibration signals are recorded and analyzed in real-time. Mel-kepstral feature parameters and line spectral frequency values are extracted from the vibration signals. Feature parameters are classified using a Support Vector Machine (SVM) which was trained a priori using a manually classified data set. An average classification rate of 96:3% and 98:3%was achieved with Antepstyle Turkish pistachio nuts and hazelnuts. An important feature of the method is that it is easily trainable for other kinds of pistachio nuts and other nuts including walnuts. © 2013 IEEE.