In disulfide rich peptides such as cyclotides and conotoxins, the peptide main chain is cross-linked by cystines. The cystine side chain is comprised of five dihedral angles that adopt certain configurations based on different combinations of these angles. The connectivity between cystine residues and configuration of the resultant side chain is essential to the activity and structure of a peptide. With their constrained nature and small size cystine rich peptides are ideal candidates for structural resolution by 2D-NMR spectroscopy. Using current methods of NMR for peptides, the prediction of the cystine connectivity is dependent on interpretation of NOE cross peaks that are often indefinite and hindered by overlap. Additionally only limited information can be derived about the side chain conformation, leading to ambiguity in the 3D-structures. This limits the ability to develop key structure activity relationships for rational drug design. It was hypothesised that there was a correlation between cystine chemical shifts and the configuration of the side chain and that if the configuration was known the connectivity could subsequently be predicted. A unique cystine database was generated that combined both NMR and X-ray structures and correlated chemical shift and structural features. Using this database, computational predictive methods such as support vector machines were developed for prediction of the configuration and connectivity. The program was able to successfully predict χ2 angles with 90% accuracy and the χ1 angles with 87% accuracy, an improvement over existing methods. It was highlighted how these extra restraints could enhance the quality of existing 3D-structures, reducing ambiguity in the backbone. Based on these results SVMs were expanded to the prediction of native cystine connectivity with a high accuracy. From these methods we aim to improve the precision of peptide solution structures determined by NMR spectroscopy, which is essential for rational development of peptide drug candidates.