Sphingolipid and cholesterol-rich liquid ordered lipid domains (lipid rafts) have been studied in both eukaryotic cells and model membranes. However, while the coexistence of ordered and disordered liquid phases can now be easily demonstrated in model membranes, the situation in cell membranes remains ambiguous. Unlike the usual situation in model membranes, under most conditions, cell membranes rich in sphingolipid and cholesterol may have a "granular" organization in which the size of ordered and/or disordered domains is extremely small and domains may be of borderline stability. This review attempts to explain the origin of the divergence between of our understanding of rafts in model membranes and in cells, and how the physical properties of model membranes can help explain many of the ambiguities concerning raft formation and properties in cells. How physical principles of ordered domain formation relate to limitations of detergent insolubility and cholesterol depletion methods used to infer the presence of rafts in cells is also discussed. Possible modifications of these techniques that may increase their reliability are considered. It will be necessary to study model membrane systems more closely approximating cell membranes in order gain a complete understanding of raft properties in cells. Very high concentrations of membrane cholesterol and proteins may explain key physical characteristics of domains in cellular membranes, and are the two of the most obvious factors requiring additional study.