Comparison of new neutron and old X-ray diffraction data for single crystals of staurolite from Pizzo Forno yielded unique answers to some, but not all, outstanding structural questions. Neutron data were collected at 13(1) K for a crystal with assumed composition Li 0.07Mg 0.87Ti 0.14V 0.01Cr 0.01Mn 0.04Fe 2+ 3.00 Fe 3+ 0.06Co 0.01Zn 0.05Al 17.69Si 7.67O 48H 3.41F 0.01, M w = 1671, a = 7.8639(10), b = 16.625(2), c = 5.651(2) Å, β = 90.015(14)°, C2 m , Z = 1, D x = 3.75 g cm −3 ; 1874 (1024 unique) reflections, λ = 1.1598(1) Å, R( F) = 3.3%. The diffraction evidence is consistent with full occupancy of the Si, Al(1), Al(2), and Al(3) sites, but not of the other ones. Detailed assignment of atoms is based on diffraction evidence and crystal-chemical arguments, but some uncertainties remain; thus exchange of (Li + Fe) by two Mg would have little effect on diffraction data. A structural model with three types of domains is proposed: ∼63% type 1, (Fe, etc.) + H(1); ∼22% type 2, (Mg, etc.) + H(2); ∼15% type 3, (Fe, etc.) without hydrogen. For the orthorhombic pseudostructure, the occupancies of the two hydrogen sites place strong restrictions on the other site occupancies. The 25(4)% observed occupancy of H(2) limits the occupancy of the nearby (Fe, etc.) site to a maximum of 75(4)%. To explain the neutron scattering, the Fe site must be occupied mainly by Fe; Li, Mn, Zn, and Mg may also occupy this site. A good ionic balance is attained for the type 2 domain if the U site from the old X-ray work is occupied simultaneously with H(2). To match the neutron data, assignment of 21(2)% Mg to the U site is plausible, but other substituents are possible. H(2) lies directly between two O(1) atoms at ∼0.9 and 2.3 Å, and H(1) is displaced away from the Fe site so that it is bonded to one O(1) at 1.01 Å and one O(3) at 2.07 Å. Four-fifths of the Fe atoms should be displaced from z = 0.25 because of electrostatic repulsion from H(1), and one-fifth should not be. This is consistent with the threefold distortion of the electron-density peak for the Fe site and the intensity ratio for the doublets in the Mössbauer spectrum. Assignment of 0.3 Al to the eight Si sites explains the low neutron scattering. Some Mg atoms, together with Ti and Fe 3+ ones, are placed in the Al sites, but assignments among the three sites are uncertain. A plausible structural formula is: [Si 7.67, Al 0.33] 8[Fe 2+ 3.00, Li 0.07, Mn 0.04, Zn 0.02, Co 0.01] 3.14[Al(1) site: Al 7.94, Fe 3+ 0.06] 8[Al(2) site: Al 7.52, Mg 0.35, Ti 0.11, V 0.01, Cr 0.01] 8[Al(3) site: Al 1.88, Mg 0.09, Ti 0.03] 2[U site: Mg 0.41, Zn 0.02] 0.43[H(1)] 2.54[H(2)] 0.86. For the monoclinic structure, the A and B subcomponents are less different for the crystal used for neutron diffraction than for the smaller one used for X-ray diffraction. Maximum deviation from the orthorhombic superstructure would be generated by the following occupancies: Al(3A) 100%, H(1A) 63%, H(2A) and Mg(A) 22%. Atoms essentially unaffected by chemical substitutions have displacement factors consistent with zero-point motions, but some (especially Fe, O(1), O(3), H(2)) have large B values indicative of more than one center-of-motion. All bond lengths and angles are reasonable when chemical substitutions are considered.