Premature termination codons (PTCs) can cause the decay of mRNAs in the nuclear fraction of mammalian cells. This enigmatic nuclear response is of interest because it suggests that translation signals do not restrict their effect to the cytoplasm, where fully assembled ribosomes reside. Here we examined the molecular mechanism for this putative nuclear response by using the T-cell receptor-beta (TCR-beta) gene, which acquires PTCs as a result of programmed rearrangements that occur during normal thymic ontogeny. We found that PTCs had little or no measurable effect on TCR-beta pre-mRNA levels, but they sharply depressed TCR-beta mature mRNA levels in the nuclear fraction of stably transfected cells. A PTC split by an intron was able to trigger the down-regulatory response, implying that PTC recognition occurs after an mRNA is at least partially spliced. However, intron deletion and addition studies demonstrated that a PTC must be followed by at least one functional (spliceable) intron to depress mRNA levels. One explanation for this downstream intron-dependence is that cytoplasmic ribosomes adjacent to nuclear pores scan mRNAs still undergoing splicing as they emerge from the nucleus. We found this explanation to be unlikely because PTCs only 8 or 10 nt upstream of a terminal intron down-regulated mRNA levels, even though this distance is too short to permit PTC recognition in the cytoplasm prior to the splicing of the downstream intron in the nucleus. Collectively, the results suggest that nonsense codon recognition may occur in the nucleus.