Cripto is a membrane-bound co-receptor for Nodal, a member of the transforming growth factor-beta superfamily. Mouse embryos lacking either Cripto or Nodal have the same lethal phenotype at embryonic day 7.5. Previous studies suggest that O-fucosylation of the epidermal growth factor-like (EGF) repeat in Cripto is essential for the facilitation of Nodal signaling. Substitution of Ala for the Thr to which O-fucose is attached led to functional inactivation of both human and mouse Cripto. However, embryos null for protein O-fucosyltransferase 1, the enzyme that adds O-fucose to EGF repeats, do not exhibit a Cripto null phenotype and die at about embryonic day 9.5. This suggested that the loss of O-fucose from the EGF repeat may not have led to the inactivation of Cripto in previous studies. Here we investigate this hypothesis and show the following: 1) protein O-fucosyltransferase 1 is indeed the enzyme that adds O-fucose to Cripto; 2) Pofut1(-/-) embryonic stem cells behave the same as Pofut1(+/+) embryonic stem cells in a Nodal signaling assay; 3) Pofut1(-/-) and Pofut1(+/+) embryoid bodies are indistinguishable in their ability to differentiate into cardiomyocytes; and 4) none of 10 amino acid substitutions at Thr(72), including Ser which acquires O-fucose, rescues the activity of mouse Cripto in Nodal signaling assays. Therefore, the Thr to which O-fucose is linked in Cripto plays a key functional role, but O-fucose at Thr(72) is not required for Cripto to function in cell-based signaling assays or in vivo. By contrast, we show that O-fucose, and not the Thr to which it is attached, is required in the ligand-binding domain of Notch1 for Notch1 signaling.