A single-column model coupled to a bulk microphysics parameterization (with prognostic cloud liquid water, cloud ice, rain, and snow mixing ratios) is evaluated using cloud properties retrieved at the Surface Heat Budget of the Arctic Ocean experiment (SHEBA) during the period of 1 April to 16 May 1998. Overall, the model accurately simulates the cloud boundaries and total cloud fraction, but has difficulty correctly partitioning the cloud phases and predicting the condensed water contents and paths. In particular, the mean liquid water path (LWP) is underestimated by 76%. This bias is attributed to underpredicting the liquid cloud fraction, that is, underpredicting the frequency of liquid- or mixed-phase clouds. The mean ice water path (IWP) is underestimated by 42%. Glaciation in the model occurs primarily through the preferential depositional growth of pristine ice initiated by deposition-condensation nucleation at the expense of liquid water, in contrast to glaciation mechanisms inferred from observations. Sensitivity tests are conducted to elucidate the relative importance of various microphysical parameters on the modeled cloud properties and processes. The liquid cloud fraction and mean LWP are most sensitive to uncertainties in the ice crystal number concentration, while the mean IWP is sensitive to several cloud ice/snow microphysical parameters, including the collection efficiency for riming and terminal fall velocities. The model evaluation is also discussed in the context of the spatial resolution and the approach to cloud scale separation. The unique spatial scales (particularly in the vertical) associated with Arctic stratiform clouds must be taken into account in order to correctly simulate the observed cloud properties.