The consumption of watermelon relies on the guarantee of a high quality product recently. In internal and external qualities of melons, maturity is the most important factor, which is hard to evaluate. The objective of this research was to develop a feasible non-destructive procedure for ripeness detection based on the feature of acoustic impulse response. Four features/parameters were examined for the correlation of four ripeness stages, including resonant frequency, damping coefficient, symmetry of wave, band magnitude (BM). F-ratio and analysis results showed the relationships were too weak for maturity sorting. Then, two new features, MFCC vector and band magnitude vector (BMV), were evaluated for sorting abilities. Results showed that the Dratio of BMV was the highest in all of the features, and after tested by trained PNN with new samples, the wellclassified percentage showed the best ability of BMV to detect the stages of ripeness.