The prognosis of prostate cancer is determined by using the Gleason grading system. This grading is done based upon the tissue pattern obtained from the tumor, after staining the biopsy with Heamatoxylin and Eosin (H&E). Presently, experienced pathologists manually grade on prostate cancers subjectively. The grading therefore depends upon the experience of the pathologists, quality of the staining and various other factors. To overcome this, an image analysis system is developed using MATLAB that can examine the biopsy image and grade it objectively. Size distribution of the sample image is utilized to recognize the pattern. The prediction is done based on the pattern of lumen, nuclei and the glandular organization in the representative areas of biopsy-image taken from a microscope. The results obtained show a remarkable accuracy and is closer to the manual grading scores. This Computer-Adaptive- Diagnosis (CAD) system may be used as a powerful adjunct for effectively diagnosing the prostate cancers and grading them.