In this paper we describe the application of a movement-based design process for digital musical instruments which led to the development of a prototype DMI named the Twister. The development is described in two parts. Firstly, we consider the design of the interface or physical controller. Following this we describe the development of a specific sonic character, mapping approach and performance. In both these parts an explicit consideration of the type of movement we would like the device to engender in performance drove the design choices. By considering these two parts separately we draw attention to two different levels at which movement might be considered in the design of DMIs; at a general level of ranges of movement in the creation of the controller and a more specific, but still quite open, level in the creation of the final instrument and a particular performance. In light of the results of this process the limitations of existing representations of movement within the DMI design discourse is discussed. Further, the utility of a movement focused design approach is discussed.