Consideration of the evidence presented in this review leads to the following conclusions: (a) Isolated MBP in aqueous solution has little ordered secondary or tertiary structure. (b) In this state, the protein can associate with a wide range of hydrophobic and amphiphilic compounds, these interactions involving limited sections of the protein. (c) The strength of binding to bilayers and the accompanying conformational changes in the protein are greatest for systems containing acidic lipids, presumably because of the involvement of ionic interactions. (d) When bound to bilayers of acidic lipids, MBP will have substantially more ordered secondary structure than it manifests in aqueous solution, and it is likely to be oligomeric (possibly hexameric). (e) MBP does affect the organization of lipid aggregates. It influences strongly the separation of bilayers in multilayers of purified lipids, and at present this must be viewed as its prime role within myelin. The greatest impediment to our understanding of MBP is the lack of an assayable biological activity. In contrast to the situation with enzymes, for example, we have no functional test for changes in protein structure or changes accompanying interactions with other molecules. Current evidence suggests that the protein has a structural role within myelin and that its own three-dimensional structure is strongly dependent on the molecules with which it is associated. If this picture is correct, studies of the isolated protein or of the protein in reconstituted lipid systems may yield, at best, a rough guide to the structure within its biological environment. Further clarification of the structure and function of MBP may have to await development of more powerful techniques for studying proteins bound to large molecular aggregates, such as lipid bilayers. The paucity of generally applicable methods is reflected in the fact that even low resolution structures are known for only a handful of intrinsic membrane proteins, and even more limited information exists for proteins associated with membrane surfaces. However, the increasing use of a combination of electron microscopy and diffraction on two-dimensional arrays of proteins formed on lipid bilayers (Henderson et al., 1990) offers the hope that it may not be too long before it will be possible to study at moderate resolution the three-dimensional structure of MBP bound to a lipid membrane.