Abstract: Mitochondria contain their own genome which codes for a small number of proteins. Most mitochondrial translation products are part of the membrane-embedded reaction centers of the respiratory chain complexes. In the yeast Saccharomyces cerevisiae, the expression of these proteins is regulated by translational activators that bind mitochondrial mRNAs, in most cases to their 5'-untranslated regions, and each mitochondrial mRNA appears to have its own translational activator(s). Recent studies showed that these translational activators can be part of feedback control loops which only permit translation if the downstream assembly of nascent translation products can occur. In several cases, the accumulation of a non-assembled protein prevents further synthesis of this protein but not translation in general. These control loops prevent the synthesis of potentially harmful assembly intermediates of the reaction centers of mitochondrial enzymes. Since such regulatory feedback loops only work if translation occurs in the compartment in which the complexes of the respiratory chain are assembled, these control mechanisms require the presence of a translation machinery in mitochondria. This might explain why eukaryotic cells maintained DNA in mitochondria during the last two billion years of evolution. This review gives an overview of the mitochondrial translation system and summarizes the current knowledge on translational activators and their role in the regulation of mitochondrial protein synthesis. This article is part of a Special Issue entitled: Protein import and quality control in mitochondria and plastids.