With the increasing amount of medical imaging data available through advances in imaging hardware, automated pattern classification will gain more importance in the near future. This study highlights technical considerations of cluster-based k-means pattern classification techniques applied to the segmentation of major tissue components in multi-contrast magnetic resonance images (mcMRI) of excised atherosclerotic plaque tissue. The stability of the k-means algorithm was assessed by varying the initial positions of the cluster centroids and the values of the convergence criterion. Variations of the criterion function with the number of clusters were explored. The segmentation results of the k-means clustering algorithm were compared to two color- classification algorithms, the first based on dynamic programming and principal analysis, and the second based on color-bi-partitioning. While the standard k-means algorithm was found to provide a robust method for the unsupervised classification of major components in mcMR images acquired of atherosclerotic plaque tissue under clinical conditions, the Two color-classification algorithms provided less accurate classification when compared with histological data.