tA non-destructive method for assessing the maturity of guava fruit was developed based on the mechan-ical properties obtained from dropped fruit impact responses. The levels of maturity were classified withcluster and discriminant analyses on the raw impact measurements and their derived indices. The numberof indices being processed was reduced with stepwise regression analysis. The accuracy of classificationwas improved using linear discriminant analysis to 76.3% with the penetrometer stiffness as a calibratorand to 84.2% with postharvest days as a calibrator. The performance shows that falling impact togetherwith adequate statistical analyses provides a promising non-destructive approach in assessing the matu-rity of guava. The non-destructive nature was validated by repeating the test on the same specimen. Thetest mechanism is mechanical and can therefore be integrated as the maturity inference engine on anautomated guava sorter.