In several mammalian transcription units, a transcription termination mechanism in which efficient termination is dependent on the presence of an intact 3' RNA processing site has been identified. The mouse beta maj-globin transcription unit is one such example, in which an intact poly(A) site is required for efficient transcription termination. It is now evident that 3' mRNA processing sites are not always processed with the same efficiency. In this study, we characterized several pre-mRNAs as substrates for the 3' mRNA processing reaction of cleavage and polyadenylation. We then determined whether poly(A) sites which vary in processing efficiency support a poly(A) site-dependent termination event. The level of processing efficiency was determined in vitro by assays measuring the efficiency of the pre-mRNA cleavage event and in vivo by the level of poly(A) site-dependent mRNA and gene product expression generated in transient transfection assays. The beta maj globin pre-mRNA is very efficiently processed. This efficient processing correlates with its function in termination assays using recombinant adenovirus termination vectors in nuclear run-on assays. When the beta maj globin poly(A) site was replaced by the L1 poly(A) site of the adenovirus major late transcription unit (Ad-ml), which is a poor processing substrate, termination efficiency decreased dramatically. When the beta maj globin poly(A) site was replaced by the Ad-ml L3 poly(A) site, which is 10- to 20-fold more efficiently processed than the Ad-ml L1 poly(A) site, termination efficiency remained high. Termination is therefore dependent on the yield of the processing event. We then tested chimeric poly(A) sites containing the L3 core AAUAAA but varied downstream GU-rich elements. The change in downstream GU-rich elements affected processing efficiency in a manner which correlated with termination efficiency. These experiments provide evidence that the efficiency of 3' processing complex formation is directly correlated to the efficiency of RNA polymerase II termination at the 3' end of a mammalian transcription unit.