Abstract: Improving sensitivity and resolution is particularly important in applications of magnetic resonance (MR) to biomolecular structure determination, where insensitive and/or dilute spins are detected, and medical imaging, where spatial-resolving contrast is crucial to early diagnosis of disease. The focus of this dissertation is the enhancement of sensitivity and resolution/contrast in nuclear magnetic resonance spectroscopy (NMR) and imaging (MRI) through the exploration, control, and utilization of spin chaos. Turbulent spin dynamics originate from the nonlinearity triggered by the joint action of radiation damping and the distant dipolar field, two reaction fields that inevitably come into play in modern MR experiments. The different dynamical stages of the nonlinear spin dynamics are analyzed using standard tools from dynamical systems theory. Transiently chaotic dynamics are shown to evolve into spatiotemporal chaos in the steady state. The transient spin dynamics also give rise to the formation of spatial patterns in MRI and may be used to reflect the underlying structural features of a sample. Experimental manifestations of the chaotic spin dynamics in NMR include sensitivity to intrinsic spin noise and experimental perturbations, leading to signal interferences and highly irreproducible measurements. The individual feedback fields may also generate unexpected, magnetization-dependent behavior, including dynamical frequency shifts due to the radiation damping feedback field. To counteract unwanted experimental complications due to spin chaos and nonlinear feedback interactions, periodic and differential control schemes based on radio-frequency pulses and/or gradients are proposed and demonstrated for various applications, including coupling of selected spin species under radiation damping and stabilizing chaotic evolution and the associated macroscopic observables. Finally, the unstable spin dynamics and their control are developed into a conceptually new approach for amplifying MRI contrast based on nonlinear feedback. The principles and general theory underlying contrast enhancement by individual and collaborative feedback interactions are described. Feedback-based contrast enhancement is then validated experimentally in applications ranging from improved lesion characterization to in vivo imaging of small animals. The integration of spin chaos, control, and amplification enables the magnetization or spins themselves to direct their own evolution, giving rise to new phenomena and approaches to designing MR experiments for sensitivity and contrast enhancement.