Earthquakes are a common natural hazard which can cause significant damage and loss of life. We currently cannot predict or prevent earthquakes since we do not understand the details of why earthquakes start and stop. Some complications arise due to the complexity in the earth. While it is easy, and useful, to simplify seismic faults as homogeneous 1D lines or 2D planes, that is not realistic. To understand how heterogeneities effect earthquake nucleation, rupture, and termination, I used large-scale laboratory fault to experimentally link stress heterogeneities due to fault geometry or direct fluid injection to rupture behavior.The first study presented here demonstrates how a localized increase in fluid pressure on a stressed fault can trigger dynamic slip on a 0.76 m PMMA laboratory fault. We compared the nucleation of dynamic events with and without fluid pressure. We conclude that rapid fluid injection into a low permeability fault increases multi-scale stress/strength heterogeneities which can initiate small seismic events that have the potential to grow large and propagate beyond the fluid pressurized region. The second study looks at the effect of on-fault background shear stress levels on fluid-triggered aseismic and seismic slip. When the fault was near critically stressed, slip initiated quickly after high pressure levels were reached. The dynamic slip event propagated far beyond the fluid pressurized region. In comparison, when the fault was far from critically stressed, dynamic slip initiated hundreds of seconds after high injection pressures were reached and this event was limited in size by the region affected by fluid pressure. The final study investigates a localized increase in normal stress due to a ”bump” on the fault. We found that when the bump prominence is small (∆σb/σn < 6), the bump was not able to stop rupture. When the bump was very prominent (∆σb/σn > 6), the bump acted both as an asperity and a barrier. The rupture stopped when the initial shear stress on the bump was low. When initial shear stress on the bump was high, rupture passed through the bump and resulted in a larger event than without a bump. Thus, a sufficiently prominent bump acted as a barrier (energy sink) when it was far from critically stressed and as an asperity (energy source) when it was near critically stressed. These studies increase our understanding of how stress heterogeneities can cause faults to start or stop slipping and if slip will be fast and cause significant shaking or remain slow and silent. Localized decreases in effective normal stress due to fluid pressure can initiate slip, sometimes seismic slip, but the background shear stress controls whether or not that slip ”runs away” and grows into a large earthquake. Localized increases in normal stress (and therefore effective normal stress) due to a bump, can terminate dynamic ruptures if the bump is prominent. Similar to decreases in effective normal stress, initial shear stress on the fault controls the final event size.