In this paper, a data mining technique, Fuzzy C Means clustering (FCM) is used to locate the optic disc in colour fundus image. Green plane from the RGB fundus image is considered in the proposed approach because it provides better contrast than the other red and blue planes. As the intensity of the optic disc is brighter than other region, the initial optic disc is identified by the brightest points which are called Region of Interest (ROI). FCM is a method of clustering which allows one piece of data to belong to two or more clusters. Before applying the FCM technique to cluster the optic disc, the initial ROI is de-noised by using median filtering technique. Results show the efficiency of the FCM to localize the optic disc in order to diagnose glaucoma.