The atomic structure of dioctahedral 2:1 phyllosilicate edge surfaces was calculated using pseudopotential planewave density functional theory. Bulk structures of pyrophyllite and ferripyrophyllite were optimized using periodic boundary conditions, after which crystal chemical methods were used to obtain initial terminations for ideal (110)- and (010)-type edge surfaces. The edge surfaces were protonated using various schemes to neutralize the surface charge, and total minimized energies were compared to identify which schemes are the most energetically favorable. The calculations show that significant surface relaxation should occur on the (110)-type faces, as well as in response to different protonation schemes on both surface types. This result is consistent with atomic force microscopy observations of phyllosilicate dissolution behavior. Bond-valence methods incorporating bond lengths from calculated structures can be used to predict intrinsic acidity constants for surface functional groups on (110)- and (010)-type edge surfaces. However, the occurrence of surface relaxation poses problems for applying current bond-valence methods. An alternative method is proposed that considers bond relaxation, and accounts for the energetics of various protonation schemes on phyllosilicate edges.