Gradient selection refers to the flows of mass and energy occurring along the steepest gradients of potentials or concentration. This is most intuitive for gravity-driven flows, but it also applies to wind and air flows due to gradients of atmospheric pressure, heat flux via conduction from hotter to colder, and chemical diffusion from higher to lower concentrations. Examples of gradient selection described here include wind and air flow patterns in the atmosphere, hillslope mass movements, topographic steering of wind in sand dunes, and karst dissolution dynamics. Hydraulic selection is a form of gradient selection applicable to fluvial systems and is outlined here in the context of the “job of the river” to move excess water, with geomorphic work of the river a byproduct of its hydrologic job. Preferential gradient flow paths generally persist in situations where flows can form and modify their pathways due to positive feedback (e.g., preferential flow paths in soil, groundwater, and surface hydrology), and tend to recur or persist in other situations (e.g., ocean currents and global wind patterns). Gradient selection is local, such that accessing the steepest gradient in the immediate vicinity may or may not enhance the flux efficiency of the larger system. Examples include stream capture and ice stream piracy.