Abstract. Dominant point detection was considered to extract a number of important points from a set of data points collected from some scientific phenomena or analytical studies. The extracted data points shall be able to reflect the original shape of the data. Here, a method to detect the dominant points is proposed using an exclusive formula which involved eigenvalues of the covariance matrix. Concept of region of support which played a vital role in dominant point detection is discussed. A few samples where the data points are regularly spaced and irregularly spaced are used to test the efficiency of the method. The graphical results are presented to show shape preservation of the dominant points to the shape of the data.