The crystal structure of human CG reveals that each subunit is a member of the superfamily of cystine-knot growth factors. Although the distribution of the cysteine residues in all the β-subunits is conserved, the conformation of the human FSH dimer differs from that of the CG/LH dimers. This suggests that the function of the cystine bonded loops in the human FSHβ-subunit may differ from that in the CGβ-subunit. To address this issue, we deleted two disulfide bonds in the FSHβ domain: cys 20–104 and cys 28–82, which correspond to the disulfide bonds 26–110 and 34–88, respectively, in the CGβ-subunit. The cys 26–110 bond is associated with the “seat-belt” region and cys 34–88 is a bond in the cystine knot. Coexpression of the wild-typeα -subunit with the FSHβ cysteine mutants in CHO cells revealed no detectable heterodimer. The FSHβ mutants were then incorporated into a single chain where the β-subunit is genetically fused to theα -subunit. In such a model, the rate-limiting subunit assembly step is by-passed and mutations that otherwise block heterodimer formation can be evaluated in terms of biological activity. Compared with the nonmutated single chain, the single-chain 28–82 mutant is secreted more slowly and its recovery is substantially reduced, whereas secretion and recovery of the 20–104 mutant was not significantly affected. The receptor binding affinity of the cys 28–82 mutant did not differ from wild-type and binding of the cys 20–104 mutant was decreased only 2-fold. The signal transduction data parallel the binding affinities, although the maximal accumulation of cAMP is less for the cys 20–104 mutant than that seen for cys 28–82 and nonmutated single-chains variants. These data support the hypothesis that the determinants for intracellular behavior and bioactivity of the gonadotropins are not the same, and that the cystine knot is a critical determinant for the formation of a stable, assembly-competent subunit. In addition, the data imply that the “seat-belt” conformation does not play a prominent role in the bioactivity of FSH.