Because of difficulties in design, fabrication, and testing, commercialisation of nano-electro-mechanical devices (NEMS) has proceeded much more slowly than for equivalent microscale devices. This thesis addresses several aspects that may provide solutions. Firstly, an improved mass production nanofabrication process based on optical lithography and sidewall transfer lithography (STL) is developed and demonstrated. The process allows more design freedom than previously demonstrated STL processes, for example allowing the combination of flexible and stiff mechanical nanostructures. Different nanostructures including single ribs, crossing ribs, and 3-way intersections are demonstrated. The fabricated devices have a minimum in-plane width of ca 200 nm defined by a silicon dioxide layer grown using thermal oxidation and an aspect ratio > 30:1. Secondly, methods of simulating complex nanoscale structures using the stiffness matrix method (SMM) are presented. The SMM tool replaces lengthy finite element simulations with accurate and fast simulations. These methods can be used to study trusses and other intricate designs and can also be augmented with additional multiphysics components to allow simple electrostatic modelling. Static and dynamic solvers are developed to study combined MEMS and NEMS devices. Different types of filters such as passband, stop band, and arrays of filters are investigated, and methods of improving frequency tuning are developed based on mass loading. Thirdly, methods are developed to model the back-scattered light from microscale and nanoscale objects in optical microscopy. It is shown that sub-wavelength features appear as black lines in bright-field microscopy that beyond a certain depth, while microscale features are observable only as their edges. Hence, NEMS and MEMS may both be perceived, even at optical wavelengths, and can be distinguished from the foreground in images via the detection of brightness valleys. Dark field microscopy is also investigated, and it is shown that nanoscale features will appear as bright lines and wider features as bright edges. Image processing algorithms based on valley detection, thresholding, and masking are developed to separate NEMS, MEMS, and background in optical microscope images.