The Dali server is a web-based tool for protein structure comparison and conservation mapping. It allows users to compare newly solved protein structures against those in the Protein Data Bank (PDB). The server consists of three parts: (i) the Dali server compares structures against the PDB, (ii) the Dali database allows browsing precomputed structural neighbourhoods, and (iii) the pairwise comparison generates suboptimal alignments for a pair of structures. The server accepts PDB identifiers or user-uploaded structures as input. Results are hyperlinked for interactive analysis, and the web interface is simple and user-friendly. The key purpose of interactive analysis is to check whether conserved residues align in multiple structural alignments and how conserved residues and ligands cluster in multiple structure superimpositions. In favorable cases, protein structure comparison can lead to evolutionary discoveries not detected by sequence analysis. The Dali server processes up to eight PDB searches in parallel, with others queued. Results are stored for two weeks. The server and database return only the best match for each PDB structure, while the pairwise comparison returns suboptimal matches. The Dali database is updated twice a year and contains precomputed structural alignments of PDB90 against the full PDB. The server aims to retrieve a list of 500 structural neighbors with the highest Z-scores. Most query structures have strong similarity to existing PDB structures. The server uses fast filters to identify a shortlist of about 100 promising candidates. If these produce strong matches, the search proceeds by walking. Otherwise, the query structure is compared with PDB90 in one versus all fashion, followed by a walk to collect matches to redundant PDB structures. The server provides a common output format and interactive analysis tools. Results include a list of structural neighbours ranked by Z-score and alignment data. The results are presented as plain text for downloading and as hypertext for interactive analysis. The default results page reports the top 500 matches to all chains in the PDB. Selected subsets of matches can be visualized as multiple sequence alignments or in multiple 3D superimpositions. The server uses Jmol for molecular graphics. Structural conservation is calculated within the selected subset of matches. The Dali server has evolved to cope with the growth of the PDB, which is now a hundred times bigger than when the server started. The server operates with a satisfactory balance between speed and sensitivity.The Dali server is a web-based tool for protein structure comparison and conservation mapping. It allows users to compare newly solved protein structures against those in the Protein Data Bank (PDB). The server consists of three parts: (i) the Dali server compares structures against the PDB, (ii) the Dali database allows browsing precomputed structural neighbourhoods, and (iii) the pairwise comparison generates suboptimal alignments for a pair of structures. The server accepts PDB identifiers or user-uploaded structures as input. Results are hyperlinked for interactive analysis, and the web interface is simple and user-friendly. The key purpose of interactive analysis is to check whether conserved residues align in multiple structural alignments and how conserved residues and ligands cluster in multiple structure superimpositions. In favorable cases, protein structure comparison can lead to evolutionary discoveries not detected by sequence analysis. The Dali server processes up to eight PDB searches in parallel, with others queued. Results are stored for two weeks. The server and database return only the best match for each PDB structure, while the pairwise comparison returns suboptimal matches. The Dali database is updated twice a year and contains precomputed structural alignments of PDB90 against the full PDB. The server aims to retrieve a list of 500 structural neighbors with the highest Z-scores. Most query structures have strong similarity to existing PDB structures. The server uses fast filters to identify a shortlist of about 100 promising candidates. If these produce strong matches, the search proceeds by walking. Otherwise, the query structure is compared with PDB90 in one versus all fashion, followed by a walk to collect matches to redundant PDB structures. The server provides a common output format and interactive analysis tools. Results include a list of structural neighbours ranked by Z-score and alignment data. The results are presented as plain text for downloading and as hypertext for interactive analysis. The default results page reports the top 500 matches to all chains in the PDB. Selected subsets of matches can be visualized as multiple sequence alignments or in multiple 3D superimpositions. The server uses Jmol for molecular graphics. Structural conservation is calculated within the selected subset of matches. The Dali server has evolved to cope with the growth of the PDB, which is now a hundred times bigger than when the server started. The server operates with a satisfactory balance between speed and sensitivity.