The proliferation of online media has sparked the development of non-linear video forms, which allow for the variation of moving image sequences according to user input or other variables. In order to explore both the technological and artistic possibilities of non-linear video, this thesis introduces the idea of generative editing for music videos, building on previous developments in generative art and metacreation. As a first step, a conceptual framework was defined comprising investigations into artistic movements in non-linear video, media studies on generative and interactive art, and an analysis of music video. These concepts served to support the subsequent development of algorithmic approaches to music video editing and their implementation and testing in Dividation, a prototype software system utilised to explore generative music video editing methods using the author’s original artwork. While most non-linear videos tend to emphasise user interactivity, the development of Dividation aimed to derive generative methods capable of creating a different sequence every time the video is viewed, without the need for user input. For this purpose, the role of automation and randomness as means to generate variability in a renewed creative process was closely examined. In addition, the possibility of formulating editing dynamics was analysed, and the potential of this creative practice was explored. Music video analyses conducted as part of this study have shown that generative editing methods need to be part of an open system allowing editors to be involved in the specification of individual progressive requirements for any given music video. While the possibility of videographic notation or any other future systematic description of editing practice has been considered, the results have also shown that editing practice is currently most successful when relying on substantial human input because music videos depend on a large number of parameters describing both musical and visual features and their changing interplay in these sequences. With these ideas in mind, the development of the generative editing methods for Dividation was based on a bottom-up approach, with the author’s editing practice and experience guiding the process.