Dear Paul, It sounds as though you have checked the raw global intensities for outliers etc, which one could do with something like: P = spm_get(inf,'img', 'Select all the images in the analysis'); V = spm_vol(P); for i = 1:length(V) g(i) = spm_global(V(i)); end g The compression effect you are seeing implies that there is a much greater range in the patient's globals than for the controls. Have you checked whether the globals correlate with the design - for the depressed patients for example? In PET, higher global counts may be associated with better signal to noise - I guess this might explain some of what you are seeing. See you, Matthew %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%