The paper introduces the Secondary-Structure Matching (SSM) algorithm, a tool for fast three-dimensional protein structure alignment. SSM uses a graph-theoretical approach to match secondary structure elements (SSEs) and then aligns the protein backbones using Cα atoms. The authors compare SSM results with other protein comparison servers and discuss the advantages and disadvantages of different scoring methods. They propose a new score that balances root mean square deviation (RMSD) and alignment length, which shows reasonable agreement among different servers. The efficiency of SSM is demonstrated through its implementation as a web server, capable of delivering high-quality alignments in less than a minute. The paper also explores the relationship between sequence identity and structural similarity, showing that sequence identity does not always correlate with structural similarity, especially at low sequence identities. The Q score, which combines RMSD and alignment length, is found to be a more reliable indicator of structural similarity, particularly for less similar structures.The paper introduces the Secondary-Structure Matching (SSM) algorithm, a tool for fast three-dimensional protein structure alignment. SSM uses a graph-theoretical approach to match secondary structure elements (SSEs) and then aligns the protein backbones using Cα atoms. The authors compare SSM results with other protein comparison servers and discuss the advantages and disadvantages of different scoring methods. They propose a new score that balances root mean square deviation (RMSD) and alignment length, which shows reasonable agreement among different servers. The efficiency of SSM is demonstrated through its implementation as a web server, capable of delivering high-quality alignments in less than a minute. The paper also explores the relationship between sequence identity and structural similarity, showing that sequence identity does not always correlate with structural similarity, especially at low sequence identities. The Q score, which combines RMSD and alignment length, is found to be a more reliable indicator of structural similarity, particularly for less similar structures.