In most species the 3' splice site is recognized initially by an interaction between the two-subunit splicing factor USAF with the polypyrimidine (poly(Y)) tract that results in recruitment of the U2 snRNP to the branch-point consensus just upstream. In contrast, in Caenorhabditis elegans, both the poly(Y) tract and the branch-point consensus sequences are missing, apparently replaced by the highly conserved U(4)CAG/R 3' splice site consensus. Nevertheless C. elegans U2AF(65) is very similar to its mammalian and fly counterparts and may recognize the 3' splice site consensus. Here we report the cloning of the C. elegans U2AF(35) gene, uaf-2. We show that it lacks an identifiable RS domain, which, in flies, has been shown to play a role in RNA binding, but it contains an extended glycine-rich stretch at its C-terminus. uaf-2 is in an operon with cyp-13, a gene that encodes a cyclophilin with an RRM domain at its N-terminus. We demonstrate by RNA interference that both U2AF genes, uaf-1 (which encodes U2AF(65)) and uaf-2, are required for viability, whereas cyp-13 is apparently not.