Hi Andreas, > quick question: Is fnirt better off when superfluous neck and / or parts > of the nose / face are removed? Would you expect much of a difference? It depends. If you are using it in the "standard" way, i.e. registering to a template in standard-space and with --refmask specified to be a mask that mask out most extra-cerebral bits, then it will make no difference. If you are using it to "register some image to some other image", then it might make a difference. The execution time (and RAM requirements) will depend on the FOV of the --ref image (the --in image matters very little). It is therefore a bad idea to use your "typical structural image" with an FOV that extends almost down to the shoulders in the z-direction. It helps a lot to crop the image to be "a reasonably tight fit around the brain". Look at the MNI template to get an idea of what I mean by "reasonably tight". Good luck Jesper