Understanding how halogens are distributed among usual hydrous silicates in the lithosphere is important to constrain their deep geochemical cycle and fluid-rock interactions in subduction zones. This article presents first-principles modelling of halogen (F−, Cl−, Br−) incorporation in hydrous silicates including mica, chlorite, serpentine, amphibole, epidote and carpholite. The approach allows studying the impact of crystal chemistry on halogen partitioning by quantification of the energetic cost of halogen incorporation in minerals. Calculations are carried out in large systems where halogens are in minor to trace concentrations. Estimations show that F-bearing defects must be separated at least 9 Å from one another to reproduce trace element behaviour, this value increasing to at least 10 Å for Cl and Br. Results highlight the competition between the effects of electrostatic interactions and steric hindrance for incorporation of halogens, where steric hindrance has greater importance for heavy halogens, in particular for Br. Interaction with alkalis is a major control for F incorporation, especially in mica. Other parameters such as octahedral site occupancy, Si/Al ratio of tetrahedral sites and the nature of alkalis in amphibole and mica (K or Na) appear to play subordinate roles. Partition coefficients have been estimated in mineral assemblages in an effort to be representative of subduction zone metamorphism. Results show that pargasite, biotite and lizardite are favoured hosts for all three halogens, followed by clinochlore, tremolite and carpholite. The energetic cost of incorporating halogens into dioctahedral phyllosilicates and epidote is comparatively higher, and partitioning is predicted as unfavourable to these minerals. Fractionation between halogens in subduction zones is predicted by the evolution of mineral assemblages and partition coefficients, a consequence of the influence of crystal chemistry over halogen incorporation in hydrous silicates.