Despite numerous systematic studies, the relationships among many species within the dog family Canidae remain unresolved. Two areas of continued disagreement are the wolf-like species (in the genera Canis, Cuon,Lycaon, and Speothos) and the South American species (in 5-7 genera). Together the two groups include some 20 species, or more than half of those within the family Canidae. Previous work based on morphology, allozyme variation, and karyology have failed to provide a consistent picture of relationships among these taxa. In large part, the lack of success of the molecular analyses was due to the close relationships among many of these species. Associations among such closely related taxa are best studied with more fine-scale molecular analyses, such as those based on mitochondrial DNA sequence data. Here, it is proposed to test several specific hypotheses concerning the systematics of these taxa using phylogenetic analysis of approximately 850 base pairs of mitochondrial DNA sequence obtained by polymerase chain amplification (PCR) of genomic DNA. Because the mitochondrial DNA sequence of mammals evolves quickly (5-10 times faster than the average nuclear sequence), it is particularly appropriate for the study of closely related species. Many of these canid species have been the focus of comparative studies of carnivore behavior and thus resolution of their relationships will illuminate the potential effects of ancestry on observed patterns.