Using classical molecular dynamics simulation, we have studied the effect of edge-passivation by hydrogen (Hpassivation) and isotope mixture (with random or supperlattice distributions) on the thermal conductivity of rectangular graphene nanoribbons (GNRs) (of several nanometers in size). We find that the thermal conductivity is considerably reduced by the edge H-passivation. We also find that the isotope mixing can reduce the thermal conductivities, with the supperlattice distribution giving rise to more reduction than the random distribution. These results can be useful in nanoscale engineering of thermal transport and heat management using GNRs.Graphene 1,2 is a monolayer of graphite with a honeycomb lattice structure. It exhibits many unique properties and has drawn intense attention in the past few years. The unusual electronic properties of graphene are promising in many fundamental studies and applications, e.g., the ultrahigh electron mobility 2 and the tunable band gap and magnetic properties by the size and edge chirality of GNRs. [3][4][5][6] Graphene also has remarkable thermal properties. The measured value of thermal conductivity of graphene reaches as high as several thousand of W/m-K, 7-10 among the highest values of known materials. Previous studies [11][12][13] show that the thermal transport in GNRs depends on the edge chirality of GNRs. In realistic graphene samples, the edges are often passivated [14][15][16] and the isotope composition can be controlled. 17 Motivated by these, we study the effect of the edge H-passivation and various isotope distributions on the thermal transport in GNRs. We find that the thermal conductivity can be reduced by the edge H-passivation and tuned by the isotope distributions. Our study is useful in nanoscale control and management of thermal transport by engineering the chemical composition of GNRs.In this work, we employ the classical molecular dynamics (MD, similar to the method in Ref. 11) to calculate the thermal conductivities of GNRs. We use the Brenner potential, 18 which incorporates the many-body carbon-carbon and carbon-hydrogen interactions by introducing a fractional number of covalent bonds. This method has been successfully applied to many carbonbased systems, 19,20 especially to graphene. 11,21,22 The structures of GNRs are shown in Fig. 1 (with edge H-