We describe the results of our studies of expression of a Saccharomyces cerevisiae amber suppressor tRNALU gene (SUP53) in mammalian cells in vivo and in cell extracts in vitro. Parallel studies were carried out with the wild-type (Su-) tRNAI"U gene. Extracts from HeLa or CV1 cells transcribed both tRNALeU genes. The transcripts were processed correctly at the 5' and 3' ends and accurately spliced to produce mature tRNALu.Surprisingly, when the same tRNAI"U genes were introduced into CV1 cells, only pre-tRNAsLu were produced. The pre-tRNAsI"u made in vivo were of the same size and contained the same 5'-leader and 3'-trailer sequences as did pre-tRNAsLu made in vitro. Furthermore, the pre-tRNAsI"u made in vivo were processed to mature tRNAIeU when incubated with HeLa cell extracts. A tRNALeU gene from which the intervening sequence had been removed yielded RNAs that also were not processed at either their 5' or 3' termini. Thus, processing of pre-tRNALu in CV1 cells is blocked at the level of 5'-and 3'-end maturation. One possible explanation of the discrepancy in the results obtained in vivo and in vitro is that tRNA biosynthesis in mammalian cells involves transport of pre-tRNA from the site of its synthesis to a site or sites where processing takes place, and perhaps the yeast pre-tRNAslu synthesized in CV1 cells are not transported to the appropriate site.