A phylogeny of haemosporidian parasites (phylum Apicomplexa, family Plasmodiidae) was recovered using mitochondrial cytochrome b gene sequences from 52 species in 4 genera (Plasmodium, Hepatocystis, Haemoproteus, and Leucocytozoon), including parasite species infecting mammals, birds, and reptiles from over a wide geographic range. Leucocytozoon species emerged as an appropriate out-group for the other malarial parasites. Both parsimony and maximum-likelihood analyses produced similar phylogenetic trees. Life-history traits and parasite morphology, traditionally used as taxonomic characters, are largely phylogenetically uninformative. The Plasmodium and Hepatocystis species of mammalian hosts form 1 well-supported clade, and the Plasmodium and Haemoproteus species of birds and lizards form a second. Within this second clade, the relationships between taxa are more complex. Although jackknife support is weak, the Plasmodium of birds may form 1 clade and the Haemoproteus of birds another clade, but the parasites of lizards fall into several clusters, suggesting a more ancient and complex evolutionary history. The parasites currently placed within the genus Haemoproteus may not be monophyletic. Plasmodium falciparum of humans was not derived from an avian malarial ancestor and, except for its close sister species, P. reichenowi, is only distantly related to haemospordian parasites of all other mammals. Plasmodium is paraphyletic with respect to 2 other genera of malarial parasites, Haemoproteus and Hepatocystis. Explicit hypothesis testing supported these conclusions.