Hi! I think I found a very good solution and was wondering what people think. I changed the mask threshold in the spm defaults to 0.01 (instead of 0.8) and therefore include all voxels in the single-subject analysis. In the second-level analysis I mask my analysis with the template. Seems to work fine and the mask has the perfect size. Anything wrong with that approach? Nadine "If we knew what it was we were doing, it would not be called research, would it?" A. Einstein (1879-1955) Nadine Gaab, PhD Dept. of Psychology Stanford University 420 Jordan Hall Stanford,CA 94305-2130 e-mail: [log in to unmask] http://gablab.stanford.edu/