Dear Will and Zoltan, Thank you guys for providing excellent explanations. I appreciate it veru much. Could I ask you a second question? ;-) As you can see, I am carrying out a clinical examination, in which I compare each patient value with normal values derived from 27 controls on MRI after normalization of all patients and control images. In the previous mail, I wrote that I did not use "grand mean scaling", "Threshold masking", "threshold", "Implicit mask", "explicitly mask", "Global calculation", and "non-sphericity correction." However, I realized that this was a first-level analysis. Therefore, I thought I should have used some of them, and I tried to use them in GUI of SPM2 as followings; 1. grand mean scaling ---> scaling of overall grand mean 2. scale overall grand mean to ---> 50 3. Threshold masking ---> relative 4. threshold (relative to global) ---> 0.8 5. Implicit mask (ignore zero's)? ---> no 6. explicitly mask images? ---> no 7. Global calculation ---> mean voxel value (within per image fullmean/8 mask) 8. non-sphericity correction? ---> no Then, I obtained very similar results to those when I said no to all options described above. I am wondering which is the correct way and which values are correct if I should use grand mean scaling and masking etc. I would appreciate any help very much in advance. -- Kohkichi Hosoda M.D. Department of Neurosurgery, Hyogo Emergency Medical Center/Kobe Red Cross Hospital, 1-3-1 Wakinohama-Kaigan-Dori, Chuo-ku, Kobe, 651-0073, JAPAN Telephone 078-241-3131 Fax 078-241-2772 e-mail address : [log in to unmask]