The apocytochrome b gene, exclusively encoded by the mitochondrial genome, was engineered so that it could be expressed in the yeast cytoplasm. Different combinations of the apocytochrome b transmembrane domains were produced in the form of hybrid proteins fused to both the N-terminal mitochondrial targeting sequence of the ATPase subunit 9 from Neurospora crassa and to a cytoplasmic version of the bI4 RNA maturase, localised on the N-terminal and C-terminal sides, respectively, of the hydrophobic stretches. The bI4 RNA maturase, which can complement mitochondrial mutations, was used as an in vivo reporter to assess the mitochondrial import of the different groups of transmembrane helices. This new, reliable and sensitive reporter activity allowed us to experimentally determine the limitations to the mitochondrial import of hydrophobic proteins. All eight transmembrane helices of apocytochrome b could be imported into mitochondria, either alone or in combination, but no more than three to four transmembrane helices could be imported together at one time. This limit is close to that observed in the population of nuclear-encoded mitochondrial proteins. The hydrophobic characteristics of engineered and natural proteins targeted to the mitochondrial inner membrane revealed two factors important in the import process. These were (a) the local hydrophobicity of a transmembrane segment, and (b) the average regional hydrophobicity of the protein over an extended length of 60-80 residues. Such features may have played a major role in the evolution of mitochondrial genomes.