The geometries of clay smears produced in a series of direct shear experiments on composite blocks containing a clay-rich seal layer sandwiched between sandstone reservoir layers have been analyzed in detail. The geometries of the evolving shear zones and volume clay distributions are related back to the monitored hydraulic response, the deformation conditions, and the clay content and strength of the seal rock. The laboratory experiments were conducted under 4 to 24 MPa (580–3481 psi) fault normal effective stress, equivalent to burial depths spanning from less than approximately 0.8 to 4.2 km (0.5 to 2.6 mi) in a sedimentary basin. The sheared blocks were imaged using medical-type x-ray computed tomography (CT) imaging validated with optical photography of sawn blocks. The interpretation of CT scans was used to construct digital geomodels of clay smears and surrounding volumes from which quantitative information was obtained. The distribution patterns and thickness variations of the clay smears were found to vary considerably according to the level of stress applied during shear and to the brittleness of the seal layer. The stiffest seal layers with the lowest clay percentage formed the most segmented clay smears. Segmentation does not necessarily indicate that the fault seal was breached because wear products may maintain the seal between the individual smear segments as they form. In experiments with the seal layer formed of softer clays, a more uniform smear thickness is observed, but the average thickness of the clay smear tends to be lower than in stiffer clays. Fault drag and tapering of the seal layer are limited to a region close to the fault cutoffs. Therefore, the comparative decrease of sealing potential away from the cutoff zones differs from predictions of clay smear potential type models. Instead of showing a power-law decrease away from the cutoffs toward the midpoint of the shear zone, the clay smear thickness is either uniform, segmented, or undulating, reflecting the accumulated effects of kinematic processes other than drag. Increased normal stress improved fault sealing in the experiments mainly by increasing fault zone thickness, which led to more clay involvement in the fault zone per unit of source layer thickness. The average clay fraction of the fault zone conforms to the prediction of the shale gouge ratio (SGR) model because clay volume is essentially preserved during the deformation process. However, the hydraulic seal performance does not correlate to the clay fraction or SGR but does increase as the net clay volume in the fault zone increases. We introduce a scaled form of SGR called SSGR to account for increased clay involvement in the fault zone caused by higher stress and variable obliquity of the seal layer to the fault zone. The scaled SGR gives an improved correlation to seal performance in our samples compared to the other algorithms.