Mixed lipid membranes play a crucial role in numerous cellular processes and pharmaceutical applications. Fully understanding the interactions between membranes and biomacromolecules is not possible without gaining insight into underlying physical concepts. In this thesis we develop theoretical models that aim to rationalize a number of experimental findings, all involving lipid layers and their interaction with macromolecules. Our models are phenomenological and employ a minimal set of order parameters, thus focusing on essential physical interactions. We address four major subjects: First, certain mixed model membranes containing cholesterol are able to undergo macroscopic phase separation. Based on a previously suggested thermodynamic model we demonstrate that peripherally adsorbed membrane proteins tend to further facilitate phase separation, especially when they exhibit attractive interactions. Second, we show that the coupling between the two leaflets of a mixed lipid bilayer can influence its phase behavior. To this end, we calculate detailed phase diagrams and argue that their predictions are in principal agreement with experimental observations. Specifically, the coupling can trigger or inhibit phase separation, depending on lipid compositions in each leaflet and coupling strength. Third, we investigate the fundamental question if physiological pH-changes are sufficient -- and can this be employed by cellular processes -- to trigger the adsorption of peripheral proteins. Proposing a model for the previously suggested electrostatic-hydrogen bond switch mechanism, we show that protein adsorption based on electrostatic interactions alone has a weak pH dependence but is rendered pH sensitive by the electrostatic-hydrogen bond switch. Finally, the transfer of hydrophobic drug molecules in model systems from donor liposomes to a target carrier is known from experimental work to typically exhibit a first-order kinetics, sometimes also sigmoidal behavior. We develop a detailed kinetic model for drug transfer that is based on a statistical description of drug occupation numbers in liposomes and includes both drug diffusion and liposome collision mechanisms.