The continuity of clay smears evolving in sealed direct shear experiments of initially intact sandstone-mudrock sequences was quantified to large displacements up to more than ten times the thickness of the sealing layer. The sample blocks consisted of a preconsolidated clay-rich seal layer, which was embedded and synthetically cemented in quartz sand. The mineralogy and mechanical properties of the clay layer and the reservoir sandstones were varied systematically to mimic a range of natural clastic rock sequences. The fluid-flow response across the fault zone was monitored continuously during deformation using a new type of direct shear cell. The displacement at which seals break down is closely linked to the amount of phyllosilicates in the seal layer. Contrary to expectations, softer seal layers do not seal better than stiff seal layers for a given clay content. In the testing range of normal effective stresses between 4 to 24 MPa (580-3481 psi) covering maximum burial depth conditions of approximately 800 m (2625 ft) to approximately 4 km (2 mi) (assuming normal fault tectonics), a systematic trend is also observed, indicating better smear continuity by increasing the effective normal stress. Predominantly brittle processes such as slicing and wear, and not ductile drag or plastic flow, appear to be responsible for the generation of clay smears. The test results offer the prospect of incorporating critical shale smear factors (i.e., normalized displacement at which seal breakdown occurs) into probabilistic fault seal algorithms that consider important properties that can be measured or estimated, namely, clay content and fault-normal effective stress.