Multivariate analysis techniques which produce easily interpretable results, are very useful in the analysis of data corresponding to a large number of variables. One fundamental distinction between these many techniques is that some analyses are primarily concerned with relationships between variables (variable-directed), while others are primarily concerned with relationships between samples (individual-directed or sample-directed). Principal components analysis, the variable-directed multivariate analysis technique, aims to simplify the problem by making the data easier to understand and interpret by transforming the observed variables to a new set of variables which are uncorrelated and arranged in decreasing order of importance. However, unit scales of variables constitute the most significant aspect of the analysis, since the units expressed in different scales will give rise to extraction of different loadings representing different elliptical surfaces. Therefore, the main issue focuses on the determination of variables' unit scales which will serve towards the best fit for reliably interpretable results. In this study, to achieve this goal, the most convenient unit scales for river water quality variables were experimented and such extracted components were further evaluated for reliability purposes by using multiple regression method assuming that if the dominant variables in the individual components are really contributing to a single process or change in the data, it should be supported to an extent by the regression equations derived as well. The application suggested that the best fit of the quality variables' unit scales for the principal components analysis was the loads rather than concentrations resulting in the conclusion that observed mass concentrations in a river body cannot be evaluated without corresponding flow observations.