Fault damage zones can have a major impact on fluid flow through sub-surface reservoirs. The typical resolution of seismic reflection data is such that faults with throws <15m are not imaged, and those with throws >15 m are imaged as discrete planes, revealing none of the smaller scale architecture of the fault damage zones. Previous field studies show that damage zone width scales with fault throw, which suggests that a relationship exists between fault growth and increasing damage zone width. However, this hypothesis remains largely untested and the factors controlling damage zone evolution are poorly understood. This study develops kinematic models to describe the evolution of damage zones during fault growth. The predictions of these models are tested against quantitative geometric attributes of natural fault damage zones preserved in siliciclastic sand/shale sequences from the Carboniferous Northumberland Basin, NE England. These data, obtained from faults with throws spanning 0.1-20 m, were measured from detailed (cm-resolution) digital outcrop models captured using terrestrial laser scanning techniques. Study locations include areas of active open-cast coal mining that provide good 3D exposure of faults during progressive coal extraction. The damage zones comprise complex arrays of structural elements including: fault splays and oversteps; drag folds; rotated fault-bound blocks; sub-parallel fracture sets and ductile shear zones; cataclasite lenses; and intensely deformed scaly gouge. We propose two complimentary kinematic models to explain the structural relationships observed within these damage zones. The first model predicts the development of cataclasite lenses from fault-bounded blocks in contractional oversteps with increasing fault throw. In this scenario, the damage zone width remains approximately constant, defined by the initial fault separation. The second model describes the space incompatibility that develops between discrete fault planes in coherent sandstone layers and wider damage zones in adjacent shales where throw is distributed along sub-parallel fracture sets and ductile shear zones. In this scenario, damage zone width may increase with increasing fault throw. Alternatively, the width of the damage zone may be controlled by thickness of the rheologically weaker shale. These geologically-based models highlight the importance of bed thickness and rheology - in addition to fault throw - in controlling damage zone evolution and provide a basis for predicting the likely sizes of different damage zone elements associated with seismically-imaged faults in the subsurface.