Testing of untethered subscale models, often referred to as subscale flight testing, has traditionally had a relatively minor, yet relevant use in aeronautical research and development. As recent advances in electronics, rapid prototyping and unmanned-vehicle technologies expand its capabilities and lower its cost, this experimental method is seeing growing interest across academia and the industry. However, subscale models cannot meet all similarity conditions required for simulating full-scale flight. This leads to a variety of approaches to scaling and to other alternative applications. Through a literature review and analysis of different scaling strategies, this study presents an overall picture of how subscale flight testing has been used in recent years and synthesises its main issues and practical limitations. Results show that, while the estimation of full-scale characteristics is still an interesting application within certain flight conditions, subscale models are progressively taking a broader role as low-cost technology-testing platforms with relaxed similarity constraints. Different approaches to tackle the identified practical challenges, implemented both by the authors and by other organisations, are discussed and evaluated through flight experiments.