Antimicrobial peptides (AMPs) are defined as any peptides and proteins that kill or inhibit the growth of microorganisms. They form integral parts of the innate immune systems of almost all living organisms, and are considered to be a potential new class of antibiotics, whose use could abate the current antimicrobial resistance crisis. Their sequential composition also suggests the possibility of tailoring activity, and as such a great interest has grown around the possibility of synthetically designed AMPs. Studies have indicated that nonpolar residues, such as tryptophan, and basic residues, such as arginine, are common in AMPs and important for their activity. However beyond this, little is known about the exact sequence-feature relationships of AMPs. In this thesis I explore two different datasets of AMPs to determine exact sequence-activity relationships, one a pre-existing set and the other a complete set of peptides. Examination of the pre-existing set indicated that sequence-activity relationships are different for different microorganisms, with the positioning of basic residues being shown to be important for activity against Gram-negative bacteria, and tryptophan content as important for activity against Grampositive bacteria, eukaryotic microorganisms and erythrocytes. To uncover more specific relationships, a complete set of peptides composed of all combinations of arginine and tryptophan for sequences, ranging in length from 1 to 7 residues, was screened for antimicrobial activity against S. aureus, P. aeruginosa and C. albicans, as well as haemolytic activity. Analysis of the resulting activity landscapes revealed an optimal ratio of 3:2 tryptophan to arginine residues for antimicrobial activity against all organisms, and an optimal ratio of 4:1 for haemolytic activity. Tryptophan clustering was also shown to be important for antimicrobial activity, and an even distribution of arginine residues throughout the sequence. Activity landscape analysis suggested that landscape roughness increased dramatically from single length subsets to the complete set. Aggregation analysis showed that whilst aggregation was common at high concentrations, it was much less prevalent at lower concentrations, suggesting aggregation was reversible. Membranebinding analysis highlighted significant correlations between antimicrobial activity and mass binding to anionic membranes, suggesting that the peptides in the complete set were membrane active. Scanning electron and atomic force imaging of bacteria treated with a subset of efficacious peptides exhibited an increase in morphological deformities, further indicated this. The findings outlined in this thesis will be of use in future work to direct synthetic AMP design.