We present a technique designed aiming to improve the compression of the Edgebreaker CLERS string for large and regular meshes, where regularity is understood as the compactness of the distribution of vertex degrees. Our algorithm uses a specially designed context based coding to compress the CLERS sequence. It is exceptionally simple to implement and can easily incorporated into any existing Edgebreaker implementation which uses the Spirale Reversi algorithm for decompression. Experimental results show that our procedure is very fast (600,000 triangles per second on PIII 650MHz for decompression) and leads to compression rates which are, in most cases, superior to those previously reported for large meshes of high regularity. Key Words: Mesh, Compression, Entropy, Context Figure 1: The Edgebreaker traversal of the well-known bunny and horse models. Triangles corresponding to different symbols are shown in different colors ('C'=blue, 'R'=green, 'S'=red, &...