Termination of translation in higher organisms is a GTP-dependent process. However, in the structure of the single polypeptide chain release factor known so far (eRF1) there are no GTP binding motifs. Moreover, in prokaryotes, a GTP binding protein, RF3, stimulates translation termination. From these observations we proposed that a second eRF should exist, conferring GTP dependence for translation termination. Here, we have shown that the newly sequenced GTP binding Sup35-like protein from Xenopus laevis, termed eRF3, exhibits in vitro three important functional properties: (i) although being inactive as an eRF on its own, it greatly stimulates eRF1 activity in the presence of GTP and low concentrations of stop codons, resembling the properties of prokaryotic RF3; (ii) it binds and probably hydrolyses GTP; and (iii) it binds to eRF1. The structure of the C-domain of the X.laevis eRF3 protein is highly conserved with other Sup35-like proteins, as was also shown earlier for the eRF1 protein family. From these and our previous data, we propose that yeast Sup45 and Sup35 proteins belonging to eRF1 and eRF3 protein families respectively are also yeast termination factors. The absence of structural resemblance of eRF1 and eRF3 to prokaryotic RF1/2 and RF3 respectively, may point to the different evolutionary origin of the translation termination machinery in eukaryotes and prokaryotes. It is proposed that a quaternary complex composed of eRF1, eRF3, GTP and a stop codon of the mRNA is involved in termination of polypeptide synthesis in ribosomes.