The freeze-fracture technique is now widely used in the study of membranes, but it should be stressed that it shows internal hydrophobic planes of membranes, prepared under physical conditions far removed from those prevailing in vivo. Hence there is considerable potential for artefact. Work on the membrane lipid component, the intramembrane particles, and their aggregation under certain conditions, is reviewed in the context of fact versus artefact. Particular attention is paid to the results of complementary replica experiments, performed in the author's laboratory and elsewhere, which indicate that lipid collapse and protein particle deformation contribute to the appearance of the membrane fracture face seen in the final replica. A model, showing the effects of freeze-fracturing on membranes, is presented.