We have performed a metrological characterization of the quantum Hall resistance in a 1 µm wide graphene Hall-bar. The longitudinal resistivity in the center of the ν = ±2 quantum Hall plateaus vanishes within the measurement noise of 20 mΩ upto 2 µA. Our results show that the quantization of these plateaus is within the experimental uncertainty (15 ppm for 1.5 µA current) equal to that in conventional semiconductors. The principal limitation of the present experiments are the relatively high contact resistances in the quantum Hall regime, leading to a significantly increased noise across the voltage contacts and a heating of the sample when a high current is applied. The Hall resistance in two-dimensional electron systems (2DESs) is quantized in terms of natural constants only, R H = h/ie 2 with i an integer number [1]. Due to its high accuracy and reproducibility this quantized Hall resistance in conventional 2DESs is nowadays used as a universal resistance standard [2].Recently a new type of half-integer quantum Hall effect [3,4] was found in graphene, the purely twodimensional form of carbon [5]. Its unique electronic properties [6] (mimicking the behavior of charged chiral Dirac fermions [7,8]) allow the observation of a quantized Hall resistance up to room-temperature [9, 10], making graphene a promising candidate for a high-temperature quantum resistance standard. Although the quantized resistance in graphene around the ν = 2 plateau is generally believed to be equal to h/2e 2 , up to now it has not been shown to meet a metrological standard. In this Letter we present results of the first metrological characterization of the quantum Hall resistance in graphene. In particular, we will address the present accuracy of quantization (15 ppm) and the experimental conditions limiting this accuracy.Our sample consists of a graphene Hall-bar on a Si/SiO 2 substrate forming a charge-tunable ambipolar field-effect transistor (A-FET), where the carrier concentration can be tuned with a back-gate voltage V g [11]. In order to remove most of the surface dopants that make graphene generally strongly hole doped and limit its mobility, we have annealed the sample in-situ for several hours at 380 K prior to cooling it down slowly (∆T /∆t < 3 K/min) to the base temperature (0.35 K) of a toploading 3 He-system equipped with a 15 T superconducting magnet. After annealing, the charge neutrality point in the A-FET was situated at 5 V and the sample dis- * Electronic address: J.Giesbers@science.ru.nl † Electronic address: U.Zeitler@science.ru.nl 2 µm