The qualitative comparison of Fuzzy and k-Means segmentation, with histogram guided initialization, on tumor edema complex MR images. The accuracy of any segmentation scheme depends on its ability to distinguish different tissue classes, separately. Fuzzy K-Means is exactly the same algorithm as K-means, which is a popular clustering technique. The only difference is, instead of assigning a point exclusively to only one cluster, it can have some sort of fuzziness or overlap between two or more clusters. Hence, there is a serious pre- requisite to evaluate this ability before employing the segmentation scheme on medical images. This work FKMC evaluates the ability of FCM and k-Means to segment Gray Matter (GM), White Matter (WM), Cerebrospinal Fluid (CSF), Necrotic Focus of Glioblastoma Multiform (GBM) and the perifocal vasogenic edema from pre- processed T1 contrast axial plane MR images of tumor edema complex. The experiment reveals that FCM identifies the vasogenic edema and the white matter as a single tissue class and similarly gray matter and necrotic focus, also. K-Means is able to characterize these regions comparatively better than FCM. FCM identifies only three tissue classes whereas; k-Means identifies all the six classes. The experimental evaluation of k- Means and FCM, with histogram guided initialization is performed in Matlab. Fuzzy c-means (FCM) is a widely used unsupervised pattern recognition method for medical image segmentation. The conventional FCM algorithm and some existing variants are either sensitive to noise or prone to loss of details. In this paper, presents a modified FCM algorithm that incorporates bilateral filtering for medical image segmentation. The experimental results and quantitative analyses suggest that, compared to the conventional FCM, the proposed method improves clustering performance with higher standard of noise-resistance and detail-preservation. Medical image pixels are highly correlated, i.e. the pixels in the immediate neighbors possess similar feature data. In other words, the probability that adjacent pixels belong to the same cluster is great. Therefore, effectively using relationship of neighboring pixels can be of great aid in medical image segmentation.