Hi Rick,
3dWarp with the "-newgrid 2 2 2" option worked well, but when I used 3dfractionize, with the -template option, it again cut off part of the brain, but I do get the correct voxel counts and origins. I even tried using the TT_N27 atlas as the template instead of the other dataset volumes, but am still having the same problem. Do you think the problem is with my choice of template? Also, when I took the brain out of spm, it was already talairached, but when I converted it to afni I get a +orig. file instead of a +tlrc. Am I doing something wrong in the initial stages, or should this not matter?