We have considered the extent to which details of lectin binding directly visualized by freeze-etch electron microscopy are consistent with current concepts of ganglioside arrangement in phosphatidylcholine bilayer membranes. Native lectins in general seem appropriate labels for this type of study. Wheat germ agglutinin, Ricinus communis agglutinin, and peanut agglutinin are adequately resolved on membrane surfaces as spherical particles of diameters 6 nm, 10 nm, and 13 nm, respectively (uncorrected for platinum shadow thickness). The finite areas covered by these markers correspond to some 56, 157, and 265 lipid molecules, respectively, on the surfaces of the shadowed rigid phosphatidylcholine matrices employed here; and this constitutes a basic limitation to the precision with which one can localize a given glycolipid receptor. Ricinus communis agglutinin provides a marker whose size permits adequate quantitation of bound material while minimally obscuring detail. Using it we estimated the size limits of G M1-enriched domains, since this is the ganglioside which has shown the greatest evidence of discontinuous distribution in our hands (Peters, M.W., Mehlhorn, I.E., Barber, K.R. and Grant, C.W.M. (1984) Biochim. Biophys. Acta 778, 419–428). Results of such analyses indicate the probable existence of phase separated domains selectively enriched in G M1 up to 60 nm in extent (5600 lipid molecules) for rigid dipalmitoylphosphatidylcholine membranes bearing up to 14 mol% G M1. Similar observations were true of rigid bilayers of dimyristoylphosphatidylcholine; however, if domains enriched in G M1 exist in fluid dimyristoylphosphatidylcholine, they are on the order of 6 nm or less in diameter (or are dispersed by lectin binding). Employing the small lectin, wheat germ agglutinin, which binds to all gangliosides, we then examined the effect of exposure to Ca 2+ ions (while in the fluid state) on the ganglioside ‘domain structure’ referred to above in rigid dipalmitoylphosphatidylcholine host matrices. G M1, G D1a and G T1b were studied at 0, 2 and 10 mM Ca 2+ concentrations. It was demonstrated by spin label measurements that the dipalmitoylphosphatidylcholine matrix retained its basic melting characteristics in the presence of added Ca 2+ and ganglioside under these conditions. Within the technique's functional resolution limit of some 6 nm we were unable to identify any effect of Ca 2+ in physiological concentration on ganglioside topography as reflected by bound lectin distribution. The rigid dipalmitoylphosphatidylcholine host matrix had been selected to minimize receptor redistribution (ganglioside aggregation or disaggregation) caused by lectin probe binding or sample preparation for electron microscopy. However the above Ca 2+-related observations were basically unaltered in a matrix of intermediate fluidity and zero cooperativity obtained by the addition of 30 mol% cholesterol. In none of our samples did we see bilayer disruption that might indicate significant patches of very high local glycosphingolipid concentration.