We have employed cell-free transcription reactions with mitochondria isolated from Saccharomyces cerevisiae to study the mechanism of RNA turnover. The specificity of RNA turnover was preserved in these preparations, as were other RNA-processing reactions, including splicing, 3' end formation of mRNAs, and maturation of rRNAs. Turnover of nascent RNAs was found to occur exonucleolyticaily; endonucleolytic cleavage products were not detected during turnover of the w intron RNA, which was studied in detail. (14,43,50). The levels of other classes of RNAs are not affected in this mutant. Other nuclear mutants specifically down regulate the stability of individual mRNAs (22,46). In the case of mutations in CBP1, which destabilize the cytb mRNA (17, 19), the protein encoded by the wild-type gene apparently interacts with RNA sequences in the 5' untranslated leader of the cytb mRNA (18,35).To understand how the specificity of RNA turnover is regulated in mitochondria, it is essential to know the mechanism by which RNAs are degraded. Little is known about this process in yeast cells, for either mitochondrial or cytoplasmic RNAs (8,24). From studies with other eukaryotes and bacteria, it is clear that both endo-and 3'-exoribonuclease activities participate in mRNA decay (for a review, see references 2 and 40). In several instances, for either eukaryotic or bacterial mRNAs the initial or rate-limiting step in mRNA turnover appears to be endonucleolytic scission of the transcript at discrete sites or regions within the RNA which are UA rich (1,4,12,26). Indeed, the relative paucity of UA dinucleotides in mRNAs has been attributed to the necessity to protect mRNAs from UpAselective RNases (3). Once cleaved, mRNA fragments are then rapidly degraded, apparently by the action of 3' exoribonucleases (20,44). Although in bacteria some mRNAs decay with an overall 5'-->3' polarity (25), this directionality is apparently not due to 5' exoribonucleolytic activity, since no such enzyme has been detected in bacteria (2). Instead, progressive 5'-3' endonucleolytic cuts along the transcript with subsequent 3' exonucleolytic degradation of the released 5' terminal fragments are thought to give rise to the observed 5'-3' polarity (9, 10).