Targeting of proteins to specific organelles is essential for a wide variety of cellular functions - from signalling and trafficking to maintaining cell structure and cell defences. Cells have developed a number of different strategies to deal with the protein-targeting problem, one of which is to ‘mark’ cellular compartments by the presence of specific lipids, and the proteins or protein domains able to recognize these lipids could discern one intracellular organelle from another. Many experimental approaches have been used to study the mechanism by which protein binds to target lipid. However, for most membrane binding protein, the details of the binding process are still unclear. This is partially due to the fact that the limited ability of currently available experimental techniques to study protein-membrane interaction at atom level. In contrast, computational approaches, such as the molecular dynamics (MD) simulation technique, allow the molecular interactions in such systems to be examined in atomic detail. In this thesis, a combination of MD simulation techniques and free energy calculations have been applied into a number of membrane protein systems, in an attempt to further our understanding of membrane protein interactions. Specifically, MD simulation techniques have been used to examine the process by which the antibiotic vancomycin and Phox homology (PX) domain bind ligand both in solution and at a membrane surface. Starting from an individual vancomycin molecule plus ligand (lipid II analogues) in solution, the simulations predicted the structures of the vancomycin-ligand complex to within 0.1 nm root-mean-squared deviation from that determined experimentally. At a membrane surface, both vancomycin and the PX domain bound spontaneously to their respective ligands, lipid II and phosphoinositol. In the simulation of vancomycin, the binding of ligand was a two-step process. The initial recognition step involved the N-terminal amine group of vancomycin and the C-terminal carboxyl group of lipid II. In addition, the effect of adding membrane targeting groups on the interaction of vancomycin with charged and neutral membranes has been examined. In the simulation of PX domain, the initial recognition step involved the interaction between phosphate groups of the phosphoinositol molecule and positively charged residues located in the binding pocket of PX domain. In addition, the non-specific interactions between the β1-β2 loop and the membrane also play an important role in the interaction of membrane bound phosphoinositol molecule and PX domain. Overall, this thesis demonstrates how MD simulations and free energy calculations can be used to explain the mechanism by which proteins recognize their respective targets and demonstrates the predictive power of simulation techniques to understand these systems in detail.