Magnetic gearboxes are a developing technology that is being considered as an alternative to mechanical gearboxes for non-contact transmission of torque in numerous applications. While mechanical gearboxes are commonplace in mechanical designs due to centuries of use and refinement, they have inherent limitations or failure modes that necessitate a) significant safety factors in their specification and/or b) routine maintenance. With the non-contact transmission of torque, magnetic gearboxes have inherent overload protection and do not suffer from the wear-based failure mechanisms of mechanical gears meshing. The design of magnetic gears is focused on high torque densities, high efficiencies, high gear ratios, and low torque ripple. The pursuit of these goals comes with inherent contradictions or challenges; for example, high torque densities can lead to large radial loads, causing unacceptable deflections. High gear ratios, in turn, mean higher speeds which will increase eddy current losses and lower the overall efficiency. Successful magnetic gearing designs are an iterative process that incorporates magnetic and mechanical considerations.Over the course of two multi-stage magnetic gearbox designs, this research specifically addressed three areas of the magnetic gear design process; structural, thermal, and manufacturing/assembly. To achieve magnetic performance, laminated steel structures are commonly employed in magnetic gear designs, but these laminated structures offer lower stiffness than the solid steel alternatives and are difficult to model due to the small thickness dimension. Techniques for reducing the computational load and accurately modeling the bonding stiffness are presented and compared to experimental results. Thermal analysis of magnetic gearboxes has typically followed the lead of electrical machine researchers and used a simple conduction model for heat transfer through the air. This approach is compared to a higher fidelity model that incorporates the effect of convection in heat transfer. Both models are compared to experimental results, with the traditional approach being sufficient for most closed systems, while the convection-conduction method is suggested for systems with inlet and outlet airflows. Finally, scaled-up design, assembly, and testing of magnetic gears have identified several aspects that have to be considered when moving from smaller, benchtop devices. First, long, uninterrupted laminated stacks were found to be unrealistic due to deflections over the length of the span. Second, the much larger forces between the magnets had to be accounted for in the assembly process with special fixturing and detailed assembly processes. Finally, at a larger scale, the edge effects of the magnetic analysis cannot be ignored. This axial flux means that material choices for mechanical elements axially proximate to the magnetically active region cannot be electrically conductive or there will be significant eddy current losses.Applications of the lessons learned in this research will lead to magnetic gearbox designs that are higher performing and more economical. Such advances will make magnetic gearboxes increasingly viable as a component in energy systems and robotics, particularly in areas where access for preventative maintenance or repair is cost-prohibitive.