Scalable video coding has gained considerable attention during the past decade, due to its attractive features that efficiently support flexible transmission over heterogeneous networks and adaptive display on a wide range of devices. As coding efficiency is predominantly the governing principle of most video coding algorithms, scalable video coding thrives in incessantly improving efficiency through incorporating newly emerged technologies while preserving the scalable features. Motion scalability, being the main topic of this dissertation, is one of these contributive technologies. Motion scalability is based on the simple concept that different decoding scenarios require different motion prediction qualities in the optimized rate distortion sense. For example, lower decoding resolutions or bit rates usually demand lower motion prediction qualities in order to maintain a better balance between motion and texture coding. This concept, although simple, is not easily realizable in a practical scalable video codec. The error drifting effect introduced from quantized motion is the first problem to face, followed by the interactive issue with other scalabilities, the embedded coding of scalable motion, and the rate distortion optimized estimation algorithm for motion parameters. In this dissertation, we deal with these challenges and propose a block-based scalable motion model, which provides both motion structure and accuracy scalabilities in order to adapt to various decoding scenarios. Through the proposed model, rate distortion performance can be improved in the middle to low bit rate range. This accomplishment is jointly achieved by applying the proposed rate distortion optimized motion estimation algorithm at the encoder and the optimal motion quality selection algorithm at the bitstream extractor. Extensive simulations will be demonstrated based on a wavelet-based scalable video codec. These results verify the superiority of the proposed scalable motion solution over non-scalable ones