The Sonderforschungsbereich 160 organized this workshop to discuss and evaluate recent activities and trends in the field of membrane protein structure. Some 20 scientists were invited to present their most recent investigations in an attempt to shed some light on various facets of the structural organization of this rather elusive class of proteins. Despite all efforts there is so far not a single example where the structural determination of a membrane protein has reached the degree of comprehensiveness, accuracy and definiteness which we are eventually seeking in order to begin inquiring sensibly into the way these molecules operate. It was the central aim of the workshop to discuss how the various rapidly advancing physical and chemical approaches presently available can be combined to achieve a 'holistic' picture of the structure of a membrane protein.Ultimately it is only the atomic model which can provide us with the necessary structural information. Most advanced in this respect is bacteriorhodopsin. A model of this protein, obtained by piecing together information from various sources [ 11, was presented by R. Henderson (Cambridge). Low-dose electron microscopy combined with digital image reconstruction techniques provided the basic morphological information: The molecule is composed of seven helical rods traversing the membrane almost perpendicularly [2]. Analysis of a new two-dimensional crystal form obtained by recrystallisation[3] corroborated the boundaries of individual bacteriorhodopsin molecules in the lattice. The amino acid sequence [4] was then fitted tentatively into the morphological model. As important preliminaries the 7 helical segments and the link regions in the sequence were localized by applying the criteria of hydrophobicity and inaccessibility to proteolytic cleavage and assuming that the helices should be comparable in length. In the next step these helical segments had to be identified in the density map. To reduce the enormous number of possible assignments the following criteria were applied: (a) connectivity of non-helical link regions; (b) charge neutralization;(c) correlation of calculated electron scattering cross-sections for the individual helices with the densities determined experimentally. The model emerging as the most probable one exhibits a number of features favourable for a membrane-embedded proton pump. The buried charged amino acids are localized in the center of the molecule, possibly forming a hydrophilic channel through which protons could jump, whereas the hydrophobic, uncharged residues are directed outward towards the lipid environment. This 'inside-out' arrangement has meanwhile been confirmed by neutron diffraction experiments [5].