Hi Anderson, Good morning. It is clear, I would like to try again. Thank you so much! Best Nancy At 2016-12-13 17:17:21, "Anderson M. Winkler" <[log in to unmask]> wrote: Hi, Note however that, as inputs to randomise, these need to be cope, not thresh_zstat. All the best, Anderson On 12 December 2016 at 20:09, Nancy Li <[log in to unmask]> wrote: Hi Paul, How are you! Very gald to hear from you. Thank you so much. I wonder the first column of degine represent what, Looking forward to your confrmation. Best Nancy At 2016-12-13 04:03:05, "Paul Robinson" <[log in to unmask]> wrote: Hi, Nancy. You'll want to fslmerge -t all the thresh_zstat images into one 4d image, knowing which images are from which group (the order is important!). Then your design will look like: grp control case 1 1 0 1 1 0 1 1 0 1 1 0 1 0 1 1 0 1 1 0 1 1 0 1 1 0 1 if you have, for example, 4 controls and 5 case subjects. hth, Paul On Mon, Dec 12, 2016 at 11:55 AM, Nancy Li <[log in to unmask]> wrote: Hi, experts, I am using randomise in FSL to compare two group diffence after first level analysis in the Feat. .feat/reg_standard/stats/cope1 were treated as input image, but it showed error token. My question is which image should be the input image, thresh_Zstat1.nii? all the images should be typed separately or in one folder.Looking forward to your help, really. Thank a lot! Best Nancy