A strain gradient theory of plasticity is introduced, based on the concept of statistically stored and geometrically necessary dislocations. This theory fits within the framework of couple stress theory and involves a single material length scale, l. Minimum principles are developed for both deformation and flow theories, which reduce to conventional counterparts when l approaches zero. The theory is used to calculate size effects in plasticity, such as macroscopic strengthening due to bonded rigid particles and macroscopic softening due to voids. Constitutive potentials are derived for this purpose. Conventional plasticity theories lack a material length scale, leading to predictions that depend only on geometric lengths. However, experimental evidence suggests that material size effects exist, with smaller geometric lengths relative to a material length scale resulting in stronger plastic responses. Strain gradients, present in indentation tests and near particles, lead to enhanced hardening due to geometrically necessary dislocations. In torsion tests, strain gradients scale inversely with wire diameter, leading to stronger behavior in thinner wires. Strain hardening in metals is due to dislocation accumulation. In uniform strain fields, dislocations are stored randomly, forming dipoles that act as a forest of sessile dislocations. Geometrically necessary dislocations are stored due to strain gradients, with their density proportional to the strain gradient. The strain gradient theory models hardening from both statistically stored and geometrically necessary dislocations, with the former scaling with von Mises effective strain and the latter with a strain gradient measure and material length scale. The theory is formulated within couple stress theory, reducing to conventional J₂ deformation theory when strain gradients are negligible. Minimum principles are established for solving boundary value problems, predicting boundary layers near interfaces. The theory is extended to a J₂ flow theory version, with minimum principles in rate form. Both deformation and flow theories give identical predictions under proportional loading. The theory is applied to predict the effect of inclusion size on macroscopic response, showing that strain gradient effects have minor influence on void softening and growth but significant strengthening effects on rigid particles. The theory incorporates couple stresses, leading to higher-order partial differential equations and boundary conditions. A bimaterial interface under simple shear demonstrates the existence of boundary layers, with exponential decay lengths dependent on material properties. The theory also accounts for dislocation density variations near rigid boundaries, with strain gradient effects mimicking dislocation repulsion. The flow theory version is developed, with plastic strain rates normal to the yield surface and linear in stress rates. Minimum principles are derived for displacement and stress rates, showing consistency with conventional plasticity theories.A strain gradient theory of plasticity is introduced, based on the concept of statistically stored and geometrically necessary dislocations. This theory fits within the framework of couple stress theory and involves a single material length scale, l. Minimum principles are developed for both deformation and flow theories, which reduce to conventional counterparts when l approaches zero. The theory is used to calculate size effects in plasticity, such as macroscopic strengthening due to bonded rigid particles and macroscopic softening due to voids. Constitutive potentials are derived for this purpose. Conventional plasticity theories lack a material length scale, leading to predictions that depend only on geometric lengths. However, experimental evidence suggests that material size effects exist, with smaller geometric lengths relative to a material length scale resulting in stronger plastic responses. Strain gradients, present in indentation tests and near particles, lead to enhanced hardening due to geometrically necessary dislocations. In torsion tests, strain gradients scale inversely with wire diameter, leading to stronger behavior in thinner wires. Strain hardening in metals is due to dislocation accumulation. In uniform strain fields, dislocations are stored randomly, forming dipoles that act as a forest of sessile dislocations. Geometrically necessary dislocations are stored due to strain gradients, with their density proportional to the strain gradient. The strain gradient theory models hardening from both statistically stored and geometrically necessary dislocations, with the former scaling with von Mises effective strain and the latter with a strain gradient measure and material length scale. The theory is formulated within couple stress theory, reducing to conventional J₂ deformation theory when strain gradients are negligible. Minimum principles are established for solving boundary value problems, predicting boundary layers near interfaces. The theory is extended to a J₂ flow theory version, with minimum principles in rate form. Both deformation and flow theories give identical predictions under proportional loading. The theory is applied to predict the effect of inclusion size on macroscopic response, showing that strain gradient effects have minor influence on void softening and growth but significant strengthening effects on rigid particles. The theory incorporates couple stresses, leading to higher-order partial differential equations and boundary conditions. A bimaterial interface under simple shear demonstrates the existence of boundary layers, with exponential decay lengths dependent on material properties. The theory also accounts for dislocation density variations near rigid boundaries, with strain gradient effects mimicking dislocation repulsion. The flow theory version is developed, with plastic strain rates normal to the yield surface and linear in stress rates. Minimum principles are derived for displacement and stress rates, showing consistency with conventional plasticity theories.