One of the first things that the realignment does is to work out which voxels to use when estimating the realignment parameters. Because some voxels contribute more to the registration than others, it is possible to get a faster realignment by using only the most "useful" voxels. What it does is to select the initial set of voxels, and then remove 10% of these to see how much "information" has been lost. This continues until a certain fraction of the information remains. The plot indicates how much information remains after 10%, 19%, 27%, 34% etc of the voxels are eliminated. I think that what is happening in your case is that the voxels eliminated in the first iteration makes very little difference to the registration accuracy. When this is plotted, it results in a warning message similar to what would be expected when you type: plot([1 1],[1 1.0000000001]) As far as I can tell, I think your images should still be correctly registered. Regards, -John | at the beginning of the "realign" I got the following warning message: | | - Requested axes limit range too small; rendering with minimum range | allowed by machine precision | in spm_chi2_plot.m at line 65 | spm_realign.m (realign_series) at line 793 | spm_realign.m (realign_images) at line 661 | spm_realign.m at line 372 | | and then, the registration goes on without problems. | | What does it mean? | Are the images correctly registered? %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%