Technical systems are critical drivers of economic consumption and production, and are generally accepted to be dependent on natural systems and processes throughout their life cycle. Accordingly, their sustainability is under increasing scrutiny. However, the basic constitution of sustainability of technical systems is unclear, and views on how sustainability can be assessed and improved are inconsistent. To address these issues, the research reported in this thesis developed two generic models of technical system sustainability: the Sustainability Cycle (S-Cycle), and the Sustainability Loop (S-Loop). The general elements and relationships involved in sustainability were identified through an inductive literature investigation spanning nine sectors. Sustainability was found to constitute an ability, which is in turn an emergent property of a system and manifested to humans as behaviour that maintains something. Activities were identified as the means by which materials and energy are transformed in a system. From a sustainability perspective, the behaviour of system activities was observed to involve the production of intended output, waste, and intended resources from inputs of renewable and non-renewable resources. This behaviour is formalised in the S-Cycle model. Humans seeking improved sustainability were found to interpret the behaviour of system activities to produce knowledge, and take action on the basis of this knowledge to produce effects that alter activity behaviour. This process is formalised in the S-Loop model, which positions the S-Cycle model within the context of human knowledge and interpretations. The validity, utility, and applicability of the S-Cycle model were evaluated through: two independent worked examples; three independent industrial case studies; two expert appraisal workshops with 27 practicing engineering designers; and an analytical study of 324 sustainability performance indicators (SPIs).Through these methods, the model was applied to ten distinct technical systems and expert opinions were elicited. All model elements and relationships were supported. One additional element/relationship was identified, leading to a refined model. The model was found to be artefact independent, supporting the identification of SPIs for different technical systems, and providing a consistent view on the behaviour of different sub-systems at various levels within a technical system. The S-Loop model received a degree of support through peer review and publication in the Journal of Environmental Management. Lastly, the research and findings were critiqued, leading to the identification of advantages, disadvantages, and recommendations, and areas for future research.