HHpred is a fast server for remote protein homology detection and structure prediction, offering pairwise comparison of profile hidden Markov models (HMMs). It allows searching a wide range of databases, including PDB, SCOP, Pfam, SMART, COGs, and CDD. It accepts a single query sequence or multiple alignment as input and returns results in a user-friendly format similar to PSI-BLAST. Search options include local or global alignment and scoring secondary structure similarity. HHpred can produce pairwise query-template alignments, multiple alignments of the query with selected templates, and 3D structural models calculated by MODELLER. It provides detailed help and is accessible at http://protev.eb.tuebingen.mpg.de/hhpred. HHpred's primary aim is to provide a method for sequence database searching that is as easy to use as BLAST or PSI-BLAST but competitive in sensitivity. It offers speed, a wide range of databases, user-friendliness, flexibility, support for multi-domain proteins, comprehensive documentation, selectivity, and high sensitivity. It is among the most sensitive servers for remote homology detection and has been benchmarked against other servers in the CAFASP4 benchmark. The server uses PSI-BLAST to build an alignment of homologs for the query sequence, then generates a profile HMM from the multiple alignment. This HMM is compared with each HMM in the selected database. The database HMMs contain secondary structure information. HHsearch is used for HMM-HMM comparison, which gains sensitivity by using position-specific gap penalties. The server also allows for local or global alignment mode, with local mode being more sensitive and robust for detecting remote relationships. An example analysis of the SpoVT protein from Bacillus subtilis shows that it consists of two domains. The N-terminal domain has significant hits in SCOP, including AbrB and MazE. Despite not being classified into the same class, the alignment suggests a distant homologous relationship. The server provides a detailed summary hit list with probabilities, E-values, and alignment information. Users can edit query alignments, realign templates, and generate structural models using MODELLER. HHpred is a powerful tool for protein homology detection and structure prediction, offering high sensitivity, flexibility, and user-friendliness.HHpred is a fast server for remote protein homology detection and structure prediction, offering pairwise comparison of profile hidden Markov models (HMMs). It allows searching a wide range of databases, including PDB, SCOP, Pfam, SMART, COGs, and CDD. It accepts a single query sequence or multiple alignment as input and returns results in a user-friendly format similar to PSI-BLAST. Search options include local or global alignment and scoring secondary structure similarity. HHpred can produce pairwise query-template alignments, multiple alignments of the query with selected templates, and 3D structural models calculated by MODELLER. It provides detailed help and is accessible at http://protev.eb.tuebingen.mpg.de/hhpred. HHpred's primary aim is to provide a method for sequence database searching that is as easy to use as BLAST or PSI-BLAST but competitive in sensitivity. It offers speed, a wide range of databases, user-friendliness, flexibility, support for multi-domain proteins, comprehensive documentation, selectivity, and high sensitivity. It is among the most sensitive servers for remote homology detection and has been benchmarked against other servers in the CAFASP4 benchmark. The server uses PSI-BLAST to build an alignment of homologs for the query sequence, then generates a profile HMM from the multiple alignment. This HMM is compared with each HMM in the selected database. The database HMMs contain secondary structure information. HHsearch is used for HMM-HMM comparison, which gains sensitivity by using position-specific gap penalties. The server also allows for local or global alignment mode, with local mode being more sensitive and robust for detecting remote relationships. An example analysis of the SpoVT protein from Bacillus subtilis shows that it consists of two domains. The N-terminal domain has significant hits in SCOP, including AbrB and MazE. Despite not being classified into the same class, the alignment suggests a distant homologous relationship. The server provides a detailed summary hit list with probabilities, E-values, and alignment information. Users can edit query alignments, realign templates, and generate structural models using MODELLER. HHpred is a powerful tool for protein homology detection and structure prediction, offering high sensitivity, flexibility, and user-friendliness.