Non-antibody scaffolds are currently under investigation as a cheaper alternative to antibodies for use as therapeutics and diagnostics(1). However, engineering non-antibody scaffolds for function typically compromises protein stability via a function-stability trade off. FN3con is an ultra stable engineered fibronectin type non-antibody scaffold(2), with a melting temperature greater than 100°C(3). FN3con has already been shown to retain a high thermostability when engineered towards binding a target, with one such variant having picomolar affinity towards lysozyme with 2-fold higher stability compared with a previously engineered FNfn10 domain. Maintaining such a high thermostability suggests that further functional mutations can be explored in order to greatly increase binding affinity compared with metastable FN3 domains. To test this hypothesis we are using directed evolution using yeast surface display(4) of FN3con to achieve both thermostability and binding affinity. We are also using rational engineering of FN3con in order to improve its solubility and crystallisation behaviour such that the structures of evolved proteins can be determined by X-ray crystallography(5). Together these approaches will provide insights into the stability-function interplay during protein evolution, which can be exploited in the development of biologic therapeutics.