The aim of this thesis was to generate "artificial" antibodies against a bioactive protein, GnSAF, produced by human ovarian granulosa cells that remains unidentified after 20 years of research. The library used in this study was a synthetic single-chain antibody scFv, Tomlinson J library. The antigen for biopanning was partially purified GnSAF. Screening the antibody clones from the library incorporated an additional selection step: an in vitro rat monolayer bioassay for GnSAF based on the specific suppression of GnRH-induced LH secretion. The initial screening with a binding ELISA technique resulted in 8 clones that were tested by bioassay, initially as pooled phage forms and subsequently as individual soluble scab forms. Three scabs recognised GnSAF bioactivity; with the suppression of GnRH-induced LH secretion by GnSAF-containing preparations reduced by up to 50% following incubation with the scabs. In order to improve the stability of the scabs for immunopurification purposes, and to widen the range of secondary labelled-antibodies available, the scabs were engineered into full length human immunoglobulins (IgG). One clone of the purified IgG form significantly altered GnSAF dose-response curves and demonstrated high affinity for GnSAF bioactivity when immobilised. When used for repeated immunopurification cycles and then Western blotted, this antibody enabled the isolation of a distinct band at around 66 kDa suggesting that this might be GnSAF. The main candidate molecules identified from the immunopurified material by excision of 2-D gel protein spots was a human serum albumin precursor and alloalbumin. This study demonstrates that the combination of bioassay and phage display technologies is a powerful tool in the study of uncharacterised proteins.