Previous complementation studies with yeast bc 1 mutants, defective in subunit VII or VIII, using heterologous and hybrid subunits, suggested that the requirement for import into mitochondria might significantly restrict the scope of this test for compatible proteins. Prediction algorithms indicate that the N-terminal domain of subunit VII contains all known characteristics of a mitochondrial targeting signal, whereas in subunit VIII such a signal is absent from the N-terminal domain, but possibly present in an internal region of the protein. Despite the fact that the characteristics of a mitochondrial import signal are found in the N-terminus of all known subunit-VII orthologues, in vitro import experiments show that the protein of human origin is not imported into yeast mitochondria. In vitro import can be restored, however, by replacement of the N-terminal part of the human protein by the N-terminus of the Saccharomyces cerevisiae orthologue, indicating a requirement for speciesspecific elements.Similar experiments were performed with subunit VIII and orthologues thereof, including a hybrid protein in which the N-terminus of the bovine heart orthologue was replaced by that of S. cerevisiae. The ability of yeast mitochondria to import this hybrid protein, in contrast with the bovine subunit-VIII orthologue itself, indicates that for subunit VIII also the N-terminus, in contradiction of theoretical predictions, contributes to the targeting signal, most likely via species-specific elements.Our findings expose the limitations of the currently available criteria for prediction of the presence and location of a mitochondrial targeting sequence and highlight the necessity of performing separate import studies for interpreting complementation studies as long as the species-specific characteristics of the import signals have not been identified.