In biological processes, such as fission, fusion and trafficking, it has been shown that lipids of different shapes are sorted into regions with different membrane curvatures. This lipid sorting has been hypothesized to be due to the coupling between the membrane curvature and the lipid's spontaneous curvature, which is related to the lipid's molecular shape. On the other hand, theoretical predictions and simulations suggest that the curvature preference of lipids, due to shape alone, is weaker than that observed in biological processes. To distinguish between these different views, we have directly measured the curvature preferences of several lipids by using a fluorescence-based method. We prepared small unilamellar vesicles of different sizes with a mixture of egg-PC and a small mole fraction of N-nitrobenzoxadiazole (NBD)-labeled phospholipids or lysophospholipids of different chain lengths and saturation, and measured the NBD equilibrium distribution across the bilayer. We observed that the transverse lipid distributions depended linearly on membrane curvature, allowing us to measure the curvature coupling coefficient. Our measurements are in quantitative agreement with predictions based on earlier measurements of the spontaneous curvatures of the corresponding nonfluorescent lipids using X-ray diffraction. We show that, though some lipids have high spontaneous curvatures, they nevertheless showed weak curvature preferences because of the low values of the lipid molecular areas. The weak curvature preference implies that the asymmetric lipid distributions found in biological membranes are not likely to be driven by the spontaneous curvature of the lipids, nor are lipids discriminating sensors of membrane curvature.