Hi, see the randomise help page on the FMRIB web pages for an explanation on what the number of permutations ("n") is all about. http://www.fmrib.ox.ac.uk/fsl/randomise/index.html Basically, it is the precision you determine your p-value with that increases with n. 5000 is actually the lower limit, 10000 are "better". Cheers, Cornelius Am 06.12.2010 um 23:32 schrieb Nicoḷ Cardobi: > Good Morning again... > > Finally, I've been running correctly dual_regression script. Thanks a lot! > > A small advice... how much permutation do you suggest in order to get a proper result? > > In the Filippi paper (on the APOE carrier etc.) i see 5000 permutation. That takes a lot of time, but if it's necessary, there is no problem. > > Thank you! > > Best Regards > > Nicoḷ