This paper concerns the feasibility of using near infrared technology and multivariate data analysis for detecting taste attributes of fruits. The aim is to introduce a non-destructive system related to development of taste components for detecting the characteristic taste parameters in different commodities. The method will have potentials for online fruit grading in warehouses and public fruit markets. A NIR spectrometer, with Photo Diode Array detector, was used to predict some of the internal quality parameters ie Soluble Solid Contents (SSC) and acidity; and to sense characteristic taste parameters of two different apple varieties (Aroma and Elstar). Non-destructive prediction and classification was based on optical reflectance in NIR range (700-1100 nm). A reasonable correlation between NIR and some quality parameters was achieved. Samples having the same ratio of SSC and acidity were correctly classified in 100 and 85% of the cases, respectively, at 10% significance level. Successful classi- fication indicates that the NIR technique has a high potential for detecting taste characteristics of apples.