Stepwise multiple discriminant analysis was applied to a two year water quality monitoring study of the Buckingham Canal at Madras, India. The variables were divided into 1. physical and chemical 2. pollutants; and 3. biological, to test which group is efficient in discriminating stations which differ in water quality. Biological variables obtained higher loadings on the discriminant functions than the other two groups. The derived discriminant functions were tested for their discriminating power through the classification phase with the same raw data. The 'Hit and Misses' tables of classification indicate that 81.66% of the samples were correctly classified by the biological variables, against only 46.66% of the samples in the case of physical and chemical variables and 48.33% in the case of pollutants.