In this paper, we propose a novel optimal quality adaptation algorithm for MPEG-4 fine granular scalability (FGS)stream over wired network. Our algorithm can maximize perceptual video quality by minimizing video quality variation and increasing available bandwidth usage rate. Under the condition that the whole bandwidth evolution is known, we design an optimal algorithm to select layer. When the knowledge of future bandwidth is not available, we also develop an online algorithm based on the optimal algorithm. Simulation showed that both optimal algorithm and online algorithm can offer smoothed video quality evolution.