The recombinations and mutations that plant mitochondrial DNA has undergone during evolution have led to the inactivation or complete loss of a number of the 'native' transfer RNA genes deriving from the genome of the ancestral endosymbiont. Following sequence divergence in their genes, some native mitochondrial tRNAs are 'rescued' by editing, a post-transcriptional process which changes the RNA primary sequence. According to in vitro studies with the native mitochondrial tRNA(Phe) from potato and tRNA(His) from larch, editing is required for efficient processing. Some of the native tRNA genes which have been inactivated or lost have been replaced by tRNA genes present in plastid DNA sequences acquired by the mitochondrial genome during evolution, which raises the problem of the transcriptional regulation of tRNA genes in plant mitochondria. Finally, tRNAs for which no gene is present in the mitochondrial genome are imported from the cytosol. This process is highly specific for certain tRNAs, and it has been suggested that the cognate aminoacyl-tRNA synthetases may be responsible for this specificity. Indeed, a mutation which blocks recognition of the cytosolic Arabidopsis thaliana tRNA(Ala) by the corresponding alanyl-tRNA synthetase also prevents mitochondrial import of this tRNA in transgenic plants. Conversely, no significant mitochondrial co-import of the normally cytosol-specific tRNA(Asp) was detected in transgenic plants expressing the yeast cytosolic aspartyl-tRNA synthetase fused to a mitochondrial targeting sequence, suggesting that, although necessary, recognition by a cognate aminoacyl-tRNA synthetase might not be sufficient to allow tRNA import into plant mitochondria.