Effective experimental techniques are available to identify replication origin regions in eukaryotic cells. Genome-wide identification of the precise sequence elements that direct origin activity is however still not straightforward, even in the yeast Saccharomyces cerevisiae which has the best characterised eukaryotic replication origins. The availability of genome sequences for a series of closely related (sensu stricto) budding yeasts has allowed us to take a 'comparative genomics' approach to this problem. Since they represent functional protein-binding sites, origin sequences are conserved better than the surrounding intergenic sequence within the genomes of closely related yeasts. We describe here how phylogenetic comparison data can be used to identify candidate replication origin sequences in the S. cerevisiae genome, and how large numbers of such candidate sites can simultaneously be assayed for ability to initiate replication. Similar approaches could potentially be used to identify protein-binding sequence elements having other functions, as well as replication origin sites in other species.