A non-redundant database of 325 protein sequences of the P450s has been sub
jected to cluster analysis. Thresholds for (sub) families have been optimiz
ed to minimize the differences between the obtained clusters and nomenclatu
re adopted for the P450s. At the given thresholds, approximately 80% of the
systematic nomenclature is reproduced by the cluster analysis. The differe
nces primarily occur among the CYP4 and CYP6 families, which include cytoch
romes P450 of mammalian and insect origin. Conflicts are also encountered a
mong plant families and among P450s of Mycobacterium tuberculosis.