The combined x-ray diffraction and electron microscopic examination of myelin has provided reasonable, but not conclusive, support for its structure as a basically bimolecular leaflet of phospholipid that is partially interspersed with protein. But there is very little basis for extending this concept to biological membranes in general. There is no adequate experimental support for the specific orientation of phospholipids as proposed in the unit membrane theory or for the proposed polar nature of protein-lipid bonds, even in myelin. Membranes differ widely in chemical composition, metabolism, function, enzymatic composition, and even in their electron microscopic image. The only similarity is their general resemblance in electron micrographs, but, until more is known about the chemistry of electron microscopy, this evidence cannot be interpreted with confidence. One positive conclusion to which I have come is that much more chemical evidence must, and can, be obtained. Techniques for the isolation of membranes are improving and protein and lipid chemistry are now highly refined arts. Quantitative analysis of many different membranes is possible and the data can be related in some instances, notably bacterial plasma membranes, to calculations of surface area. Chemical and physical changes induced in membranes of widely different lipid composition by the preparatory procedures of electron microscopy can be determined directly and correlated with the electron microscopic image. Model systems can be assembled whose compositions closely resemble those of biological membranes. Membranes can be disassociated into subunits whose properties can be studied. In particular, x-ray diffraction analysis and electron microscopy by negative staining of reaggregates of lipoproteins isolated from membranes would be very informative. Perhaps most important, the problem of membrane structure must be considered in relation to the problems of membrane function and membrane biosynthesis.