Automated segmentation of different tissue regions from brain magnetic resonance (MR) imaging has a substantial impact on many computer-assisted neuro-imaging studies. Major challenges to accomplish this task emerge from limited spatial resolution, signal-to-noise ratio, and RF coil inhomogeneity. These imaging artifacts lead to fuzziness of tissue boundaries and uncertainty in MR intensity-based tissue characterization at individual image voxels. The conventional fuzzy c-means (FCM) algorithm fails to produce satisfactory results for noisy image. In this paper, we present an entropy-based FCM segmentation method that incorporates the uncertainty of classification of individual pixels within the classical framework of FCM. Furthermore, instead of Euclidean distance, we have defined the non-Euclidean distance based on Gaussian probability density function. The new segmentation method was applied to Brainweb brain MR database at varying noise and inhomogeneity, and its performance was compared with existing FCM-based algorithms. The proposed method yields superior performance over some classical state-of-the-art methods. In addition to this, we also have performed the proposed method on some in vivo human brain MR data to demonstrate its performance.