In this research work, an attempt was made to discriminate different grades of black crush-tear-curl (CTC) tea based upon their physical attributes such as color of brewed liquor, texture, size and shape of the tea granules obtained by machine vision technique. The principal component analysis (PCA) was applied over two types of data. First, tea samples with seven different quality grades but same mechanical grading and second, samples with same quality grade but nine different mechanical grades (Brokens. Fannings and Dust) were considered for analysis, respectively. The results of PCA showed that best discrimination (100%) in both types of data was given by color attributes only. Correlations among tea samples and physical attributes were determined. Based upon these results it may be concluded that color only attributes are the most significant and sufficient for quantification of tea quality whereas other physical attributes contribute so little to quality estimation that they may be ignored. (C) 2012 Elsevier Ltd. All rights reserved.