The article reviews the recent literature on profile hidden Markov models (profile HMMs) and their applications in sequence analysis. Profile HMMs transform multiple sequence alignments into position-specific scoring systems, which are useful for searching databases for remotely homologous sequences. The methods complement standard pairwise comparison techniques and have been applied to large-scale sequence analysis. Several software implementations and large libraries of profile HMMs for common protein domains are available. Profile HMMs have performed comparably to threading methods in the CASP2 structure prediction exercise. The article discusses the advantages and challenges of using profile HMMs, including the need for ad hoc scoring systems and the complexity of training algorithms. It also highlights the development of motif-based HMMs and the increasing use of profile HMMs in protein structure prediction. The article concludes by emphasizing the importance of robust methods for automated sequence classification and annotation in the context of the human genome project.The article reviews the recent literature on profile hidden Markov models (profile HMMs) and their applications in sequence analysis. Profile HMMs transform multiple sequence alignments into position-specific scoring systems, which are useful for searching databases for remotely homologous sequences. The methods complement standard pairwise comparison techniques and have been applied to large-scale sequence analysis. Several software implementations and large libraries of profile HMMs for common protein domains are available. Profile HMMs have performed comparably to threading methods in the CASP2 structure prediction exercise. The article discusses the advantages and challenges of using profile HMMs, including the need for ad hoc scoring systems and the complexity of training algorithms. It also highlights the development of motif-based HMMs and the increasing use of profile HMMs in protein structure prediction. The article concludes by emphasizing the importance of robust methods for automated sequence classification and annotation in the context of the human genome project.