The scalability extension of H.264/AVC (H. Schwarz, et al., 2004) uses a layered approach for providing spatial, temporal, and SNR scalability. Due to this concept, only a restricted set of spatio-temporal-SNR points can be extracted and decoded from a global scalable bit-stream, and this set of points is defined by the chosen encoder configuration. In this paper, we present a new approach for providing flexible combined spatial, temporal, and SNR scalability. The increased flexibility is achieved by introducing NAL units that represent a refinement signal for a picture in a coarse-to-fine-description and can be truncated at any arbitrary point. The simulation results show that this approach is capable of providing flexible combined scalability while the coding efficiency is only slightly worse than that of the layered approach.