Hi Karsten Prior SPM12, you had cluster level based on size and voxel level based on strength, inferences From now on, all that is considered are topological features, cluster size or height (which the strength / Max voxel) so somehow it's not that different but inference wise it makes sense So the p values are the p values associated with peak height or cluster size using RFT. Hope this makes sense. Cyril Get Outlook for Android<https://aka.ms/ghei36> From: Karsten Mueller Sent: Tuesday, 19 December, 4:51 pm Subject: [SPM] peak-level To: [log in to unmask] Dear Colleagues, I have a short question about the p-values (FDR/FWE) corrected at "peak-level", because I am a bit confused about the term "peak-level" as this suggests that the p-values are corrected for the "peaks" and not for all voxels/resels. In the literature, the corrected p-values at "peak-level" are often also called corrected p-values at "voxel level", or "voxel inference". What is the right term, and why its called "peak-level" in SPM? Do they correct for the local maxima ("peaks"), only? best wishes, Karsten