We have mapped a signal sequence for mRNA 3'-end formation in Saccharomyces cerevisiae by using a Drosophila melanogaster DNA segment that complements a yeast adenine-8 mutation. That the 3' end of the transcript in S. cerevisiae nearly coincides with that in D. melanogaster is consistent with the possibility that mRNA termini are similarly determined in both organisms. Deletion analysis reveals that the complete signal is no more than 21 base pairs long. Part of the signal is the sequence TTTTTATA, which is seen in the termination region of several yeast genes. TTTTTATA appears to be able to act autonomously as a partial termination signal. The efficiency of the complete signal is affected by substitution of sequences downstream from it. This modulation of the effect of a signal is consistent with termination in S. cerevisiae, resembling rhodependent termination in bacteria.Until recently, transcription termination in eucaryotes has been poorly understood, in part because of the rapid processing that occurs at the 3' ends of transcripts for proteincoding genes. Processing usually involves cleavage of a precursor followed by polyadenylation in higher eucaryotes (6, 21). Histone mRNA 3' ends which lack polyadenylic acid were once thought to result from transcription termination (3) but now appear to be generated by processing (2, 7). The biochemical elucidation of mRNA 3' processing should be possible with the recent development of a cell-free system (23). In contrast to processing, actual transcription termination, which can occur at a considerable distance downstream from the mRNA 3' end (6, 27), has been relatively refractory to analysis.In Saccharomyces cereisisiae, polyadenylic acid addition and transcription termination appear to be coupled, as essentially all transcripts from protein-coding genes are polyadenylated [poly(A)+] (29). Therefore, the strong possibility exists that the study of 3'-end formation of an mRNA in yeast cells will shed light on the actual transcription termination event. Termination is known to be important in the regulation of a variety of bacterial operons (reviewed in references 17 and 24) and appears to be involved in the regulation of at least one eucaryotic gene (9, 10, 28). We have been investigating the sequence determinants of transcription termination on a DNA segment in S. cerev,isiae. This segment is derived from Drosophila melanogaster and complements a yeast adenine-8 mutation (15). We showed previously that mRNA 3' ends actually occur 50 to 90 base pairs (bp) downstream from a control signal, the 3' boundary of which we mapped precisely (13). Here we show that this signal is no more than 21 bp long. We also present evidence suggesting that an 8-bp portion of this signal promotes partial termination. It