This work investigates the mechanisms of binding of T cell receptors (TCRs) to Class I MHC-peptide complexes (pMHC). The structure of a TCR specific for the Melan-A tumour antigen bound to its cognate pMHC was solved to a resolution of 2.5 Å which gives insight into how this TCR could be mutated to optimize binding and subsequently used as a cancer vaccine. Detailed sequence and geometric analyses of all currently available structures of Class I TCR-pMHC complexes revealed that TCRs can bind to pMHC with a range of orientations, yet always focus on the central portion of the peptide and use a specific subset of six residues on the MHC helices for binding. The most striking finding was the use of aromatic residues in the TCR CDR loops to bind to residue Q155 on the MHC α2 helix. Attempts were also made to express and purify Toll-like receptors (TLRs) with the aim of solving one or more of these structures. However, despite testing of over 50 different constructs from 12 different TLRs or associated proteins, insufficient soluble protein expression was obtained for crystallization trials. Finally, a protein disorder prediction tool was developed to aid construct design for structural biology studies and improve the chances of obtaining protein crystals. This tool is based on a novel type of neural network and blind tests comparing it to 8 other disorder prediction tools showed it is one of the best in the field. It is freely available at www.strubi.ox.ac.uk/RONN. Analysis of large datasets revealed that the position of order/disorder transitions is quite precisely defined in amino-acid sequences and that transition regions have an amino acid composition distinct from that of bulk ordered and disordered sequences. There is a steady decrease in order-promoting residues on the ordered side of boundaries as well as a weak sequence signal, both of which signify the approaching disorder and may prove useful for improving existing disorder prediction tools.