Current climate models generally underpredict the surface concentration of black carbon (BC) in the Arctic due to the uncertainties associated with emissions, transport, and removal. This bias is also present in the Community Atmosphere Model version 5.1 (CAM5). In this study, we investigate the uncertainty of Arctic BC due to transport processes simulated by CAM5 by configuring the model to run in an “off-line mode” in which the large-scale circulation features are prescribed. We compare the simulated BC transport when the off-line model is driven by the meteorology predicted by the standard free-running CAM5 with simulations where the meteorology is constrained to agree with reanalysis products. Some circulation biases are apparent: the free-running CAM5 produces about 50% less transient eddy transport of BC than the reanalysis-driven simulations, which may be attributed to the coarse model resolution insufficient to represent eddies. Our analysis shows that the free-running CAM5 reasonably captures the essence of the Arctic Oscillation (AO), but some discernable differences in the spatial pattern of the AO between the free-running CAM5 and the reanalysis-driven simulations result in significantly different AO modulation of BC transport over northeast Asia and eastern Europe. Nevertheless, we find that the overall climatological circulation patterns simulated by the free-running CAM5 generally resemble those from the reanalysis products, and BC transport is very similar in both simulation sets. Therefore, the simulated circulation features regulating the long-range BC transport are unlikely the most important cause of the large underprediction of surface BC concentration in the Arctic.