Hello, If you lower the cluster-threshold then it will proportionally increase the size of “random” clusters relative to any real effect, and so make rejection of the null less likely. Kind Regards Matthew -------------------------------- Dr Matthew Webster FMRIB Centre John Radcliffe Hospital University of Oxford On 14 May 2018, at 08:42, M del Mar Velasco <[log in to unmask]<mailto:[log in to unmask]>> wrote: Dear all, I have performed a test using cluster-size thresholding with threshold -c 2.5. Then, I have repeated the same test with -c 1.5. I don't understand why I get significative clusters with -c 2.5 and I don't get any significance with -c 1.5. Does it make sense? Thanks, M del Mar