This thesis concerns the interaction of spin polarized electrons with the local magnetic moments in nanopatterned metallic systems. We study novel magnetic phenomena appearing in patterned thin film magnetic wires with length scales in the nanometer regime and in magnetic multilayers. The work has three mayor foci. The first is the interaction between magnetic domain walls and conduction electrons in single layer nanowires. We demonstrate the effect of using small constrictions as artificial traps for domain walls and use these structures to measure the contribution of a domain wall to the electrical resistivity. These measurements are correlated with the specific micromagnetic distribution induced by the constriction geometry. Similarly, we demonstrate and characterize the effect of spin current induced magnetization reversal in nanowires. This includes a measurement of the critical current/field phase space boundary between static and moving walls and an estimation of the intrinsic wall mobility. The second is focused on understanding the effects of spin currents on magnetoresistance and domain wall motion, in a multilayer nanostructure device exhibiting giant magnetoresistance (GMR). To demonstrate a potential application, we incorporate the effects of domain wall trapping and spin current induced domain wall motion into a nanometer scale spin-valve device. The device can be fully controlled through current and exhibits significant GMR response. This approach may be useful as a memory element in magnetoresistive random access memory (MRAM) technology, and the device serves as a proof of concept. The third focus is the understanding of the effect strain on the resistance of antiferromagnetically (AF) coupled giant magnetoresistive (GMR) multilayers containing highly magnetostrictive materials. Our measurements reveal that inverse magnetostriction effects lead to enhanced strain sensitivity in comparison to films made of the materials that compose the multilayer. A simple phenomenological model describing the measured field dependence of these effects is used to identify field-biasing values that optimize amplitude, linearity and reversibility of the effect.