We propose an optimized coding system for progressive image transmission. This system includes a robust image compression algorithm based on the SPIHT (set partitioning in hierarchical trees) coder. Previous work has shown that with a small amount of extra redundancy, we can integrate error detection into arithmetic coding. This technique is used to provide error detection and to improve the channel decoder performance (list Viterbi algorithm: LVA) by means of a joint source channel decoding scheme. Moreover, an unequal error protection (UEP) method which applies error detection coding only to bits affected by error propagation is introduced. Simulations on binary symmetric channels with bit error probabilities of 1.2×10-1 and 2×10-2, for the 512×512 Lena image at a source coding bit rate of 1 bpp, show that significant improvement in terms of packet recovery rates and PSNR can be obtained with the proposed system in comparison with the well-known system presented by Sherwood and Zeger (1997).