Improved nondestructive techniques for classification fruit during storage could be an efficient way to quality assessment of stock in the fruit trading. Fresh apple gradually deteriorates and becomes soft and dry during storage. During two months storage at 6.2°C and 20.4% relative humidity, the average firmness loss was obtained 29.14% and 32.02% for Golden Delicious and Red Delicious, respectively. Therefore, the potential of acoustic impulse response for non-destructive classification of apple fruits of different storage duration was examined. Golden Delicious and Red Delicious apples were classified using artificial neural network. Ten features of the sound impulse response of apples excited with a light mechanical impact on the equator of samples were extracted. The features used in classification of apples were the five first amplitudes and frequencies corresponding to these amplitudes. Based on exhaustive search method, different feature vectors including two, three, four and five features were also tested to find out the best feature vector combination for an optimal classification success. The feature vector including five features produced better classification results in general compared to other feature vectors for both Golden Delicious and Red Delicious apples. According to the result, five-featured vectors provide the highest F1-score of 84.9% and 84.7% for Golden Delicious and Red Delicious, respectively. The results indicated that acoustic impulse response method was potentially useful for classifying of apples according to duration of storage, but the classification accuracies need to be improved.