Landslides display a wide variety of behaviors ranging from slow persistent motion to rapid acceleration and catastrophic failure. Given the variety of possible behaviors, improvements to our understanding of landslide mechanics are critical for accurate predictions of landslide dynamics. Recent advances in remote sensing techniques, like satellite radar interferometry (InSAR), now enable high-resolution spatial and temporal measurements that provide insight into the mechanisms that control landslide behavior. In this dissertation, I use InSAR and high-resolution topographic data to identify 50 slow-moving landslides in the Northern California Coast Ranges and monitor their kinematics over 4 years. These landslides have similar mechanical properties and are subject to the same external forcings, which allows me to explore geometrical controls on kinematics. Each landslide displays distinct kinematic zones with different mean velocities that remain spatially fixed. Because these deformation patterns are sensitive to subsurface geometry, I employ a mathematical model to infer landslide thickness and find that these landslides exhibit a highly variable thickness and an irregular basal sliding surface. Time series analysis reveals that each landslide displays well-defined seasonal velocity changes with a periodicity of ∼ 1 year. These velocity variations are driven by precipitation- induced changes in pore-water pressure that lag the onset of rainfall by up to 40 days. Despite significant variations in geometry, I find no systematic differences in seasonal landslide behavior. To further explore how stress perturbations control landslide motion, I develop a mechanical model that reproduces both the displacement patterns observed at slow-moving landslides and the acceleration towards failure exhibited by catastrophic events. I find that catastrophic failure can only occur when the slip surface is characterized by rate-weakening friction and its spatial dimensions exceed a critical nucleation length that is shorter for higher effective stresses. These model simulations support my conclusions from the remote sensing analysis but also provide insight into the long-term evolution of landslides. This dissertation includes both previously published and unpublished co- authored material.