Membrane proteins play critical physiological roles and make up the majority of drug targets. Due to their generally low expression levels and amphipathic nature, membrane proteins represent challenging molecular entities for biophysical study. Mass spectrometry offers several sensitive approaches to study the biophysics of membrane proteins. By preserving noncovalent interactions in the gas phase and using collisional activation to remove solubilization agents inside the mass spectrometer, native mass spectrometry (MS) is capable of studying isolated assemblies that would be insoluble in aqueous solution, such as membrane protein oligomers and protein-lipid complexes. Conventional methods use detergent to solubilize the protein prior to electrospray ionization. Gas-phase activation inside the mass spectrometer removes the detergent to yield the isolated proteins with bound ligands. This approach has proven highly successful for ionizing membrane proteins. With the appropriate choice of detergents, membrane proteins with bound lipid species can be observed, which allows characterization of protein-lipid interactions. However, detergents have several limitations. They do not necessarily replicate the native lipid bilayer environment, and only a small number of protein-lipid interactions can be resolved. In this Account, we summarize the development of different membrane mimetics as cassettes for MS analysis of membrane proteins. Examples include amphipols, bicelles, and picodiscs with a special emphasis on lipoprotein nanodiscs. Polydispersity and heterogeneity of the membrane mimetic cassette is a critical issue for study by MS. Ever more complex data sets consisting of overlapping protein charge states and multiple lipid-bound entities have required development of new computational, theoretical, and experimental approaches to interpret both mass and ion mobility spectra. We will present the rationale and limitations of these approaches. Starting with the early work on empty nanodiscs, we chart developments that culminate in recent high-resolution studies of membrane protein-lipid complexes ejected from nanodiscs. For the latter, increasing collision energies allowed progressive removal of nanodisc components, beginning with the scaffold proteins and continuing through successive shells of lipids, allowing direct characterization of the stoichiometry of the annular lipid belt that surrounds the membrane protein. We consider future directions for the study of membrane proteins in membrane mimetics, including the development of mixed lipid systems and native bilayer environments. Unambiguous assignment of these heterogeneous systems will rely heavily upon further enhancements in both data analysis protocols and instrumental resolution. We anticipate that these developments will provide new insights into the factors that control dynamic protein-lipid interactions in a variety of tailored and natural lipid environments.