Color constancy is a phenomenon of perceiving the same color regardless of a scene illuminant. In the computational world, this is achieved by estimating the illuminant chromaticity and then correcting the captured image using a color correction technique. Already, various statistic- and learning-based unitary algorithms were proposed to estimate illuminant chromaticity. The computational time of the statistic-based methods varies with the statistical assumptions used and most of the learning-based methods have a very high computational time. In this article, we analyze the statistical results and computational time of different unitary methods on a benchmark data set to identify which method can estimate the illuminant chromaticity within permissible angular error at less computational time. This study indicates that there is no unique algorithm which can be considered to perform well on images with different settings and scenes. These findings motivated us to formulate a selection-based illuminant estimation method which chooses the fastest and optimum illuminant estimation method for the given image. In order to achieve this, the proposed method uses a classification-based hierarchical approach to organize the selection of a suitable algorithm based on computational time and performance. Therefore, the fastest unitary method at the very top, followed by remaining unitary methods are arranged in ascending order of computational time. This allows the illuminant estimation of the majority of images to be solved by using the fastest unitary method and remaining using complex methods which can perform well on a wide range of images. Experimental results on real world data set clearly demonstrate the effectiveness of hierarchical classification–based strategy in illuminant estimation.