Spatial and temporal characteristics of snow and ice thickness were determined for data, acquired during five cruises into the Bellingshausen/Amundsen and Ross Seas between September 1993 and 1995. When sorted into distance classes from the ice edge, snow and ice thicknesses show a pattern of spatial and temporal variability that reflects the relative regional importance of different sea ice growth mechanisms, motion and deformation as well as of oceanographic processes. The potential for snow ice formation was directly investigated for the drilling data. Flooding, measured prior to drilling, has a strong correspondence with negative freeboard, while moist snow shows a lesser correspondence. Wetness and flooding are strongly correlated with the ratio of snow load and ice thickness while the relation to the amount of ridging is less conclusive. Using an isostatic argument, the contribution of snow ice formation to the overall ice growth was shown to be considerably higher than that of congelation ice growth. It was found that one length unit of snow accumulation or bottom ice melt on an ice floe leads to approximately the same amount of subsidence for the floe. Further, rates of subsidence, due to snow load and ice melt, appear to be similar to uplift from congelation ice growth. The degree of isostatic balance on different spatial scales was evaluated by introducing a generalized density-weighted snow/ice thickness ratio (GSI) for the drilling data. The GSI includes slush densities in order to account for the suggested importance of snow ice formation. Deviations from isostatic balance occur locally but decrease quickly on spatial averaging scales of a few meters. The closest fit to isostatic equilibrium is reached by choosing a maximum value for the slush density. ¿ 1998 American Geophysical Union