The sensitivity of a regional climate model to lateral, upper, and lower dynamical forcing is evaluated for a pan-Arctic domain. The experiments focus on a month-long case study of May 1990, but the results are shown to be extendable to January and July 1990. Although the control simulation is found to adequately represent major tropospheric circulation features, a large negative geopotential height bias of up to 700 m is found at the upper levels of the model. Experiments find these biases are remedied through the replacement of a reflective rigid lid upper boundary condition with a local wave-absorbing layer. Geopotential height biases are reduced to a maximum underestimation of only 50 m at the upper boundary. Substantial improvement is also found in the simulated temperature and zonal wind profiles, transient heat and momentum fluxes, and stationary eddies as seen through the Eliassen-Palm flux divergence. With this wave-absorbing upper boundary condition in place, additional improvement is found using a gravity wave drag parameterization appropriate for this domain's 100 km horizontal resolution. Additional sensitivity to lateral forcing is examined, and the relative role of the upper boundary condition to skillful simulations within the Arctic basin domain is discussed. ¿ 2000 American Geophysical Union