Graphene’s remarkably large specific surface area (~ 2500 m2/g) has led to speculation that graphene supercapacitors could have specific capacitance as high as ~ 550 F/g. In practice, however, a number of factors reduce the specific capacitance of graphene-based electrodes. It is an ongoing challenge to untangle and quantify the various factors affecting the specific capacitance of graphene-based electrodes. In our work, we show that the capacitance per unit area of the graphene-liquid interface is much smaller than previously assumed. We use single-layer graphene on a flat, insulating surface to achieve well-defined surface area. Charge density in the graphene is precisely determined via Hall-effect transport measurements. For a variety of electrolytes, including 1 M Na2SO4 and the ionic liquid BMIM-PF6, the capacitance is less than 0.05 F/m2, i.e. much less than the common benchmark value 0.2 F/m2 associated with bulk metals in contact with electrolytes. We interpret our results in terms of quantum capacitance and double-layer capacitance. Combining our experimental results and capacitance model we make realistic projections for the ultimate performance of graphene-based supercapacitors.