The relationship between cloudiness and sea ice thickness over the Arctic Ocean is examined using an idealized equilibrium energy budget model and a single-column model (SCM) forced by incoming solar radiation, atmospheric poleward energy flux, oceanic heat flux, and sea ice divergence. The sensitivity of ice thickness to cloud perturbations in the SCM is qualitatively different from previous modeling studies, which did not allow the atmospheric temperature profile to respond to perturbations in cloudiness. It is shown that this restriction implicitly changes the poleward energy flux in model experiments and obscures the intrinsic effect of clouds on the arctic climate. In the present model, equilibrium ice thickness has a strong negative correlation with cloud radiative forcing at the top of the atmosphere and no correlation with cloud radiative forcing at the surface. The results demonstrate how local processes, such as the surface albedo feedback, affect the sensitivity of ice thickness to cloud perturbations and reduce its predictability. Exploratory model experiments suggest that interactions between the Arctic and the lower latitudes (via the poleward energy flux) could strongly influence the dependence of ice thickness on cloudiness. The impact of these interactions in the model was predicted qualitatively by considering how the radiative effect of clouds is partitioned between the atmosphere and the surface. The results demonstrate the importance of using cloud radiative forcing at both the surface and the top of the atmosphere when estimating the climatic effect of perturbation in cloudiness. ¿ 2000 American Geophysical Union