Abstract The distribution of fault rocks interpreted across a modelled fault in oil or gas reservoirs is most often described by its clay content derived from standard industry algorithms such as shale gouge ratio and clay smear factor. These distributions are below the mapping resolution in seismic data, and the actual processes and mechanisms for the fault-rock development are not well understood. A well-exposed and well-preserved low-throw fault in an old railroad tunnel near Salina, Utah provides the access and scale to interpret fault-rock distributions, measure their clay contents and describe the fault-rock development. A significant number of fault-rock elemental compositions were measured quickly on the outcrop surface using a hand-held X-ray fluorescence (XRF) elemental analyser with a novel surface preparation and analysis strategy. The elemental data were converted to clay contents using a small set of samples where elemental composition was calibrated to X-ray diffraction mineralogy. The mineralogy data provide a basis for evaluating the degree of mixing of protolith beds during fault-rock development in the fault core. The fault core is not randomly mixed fragments derived from the protolith (gouge or breccia), but rather discrete, thin layers parallel to the fault surface, many of which can be traced back to a source sandstone or mudstone bed. The mineralogical composition of some fault-rock layers are unchanged from their protolith source bed, but other layers are mechanical mixtures of several source beds. The shale gouge ratio algorithm under-represents the average measured fault-rock clay content. The clay smear algorithm more accurately describes the clay content distribution, but underestimates the clay content heterogeneity along the smear length. A key uncertainty for predicting fault sealing remains prediction of the lengths and continuity of smears.