Recent theoretical advances have lead to a rheology for sea ice and an increased understanding of how stresses are generated in an ice floe by thermally induced strain rates. One of the underlying assumptions of the ice rheology is that floes in general (order of 300 m or greater in diameter) are too large to bend or twist in response to vertical or horizontal variations in strain rates. In essence, there are no nonzero bending moments. A numerical model based on this rheology has been used to produce good predictions of observed ice stresses and times of ice fracturing (Lewis, 1994; Lewis et al., 1994). On the basis of these results this paper presents a conceptual model of how existing flaws in an ice floe can impact the stress state of the ice while still asuming no bending or twisting of the floe. A modified ice rheologyis developed which quantifies the extent of existing cracks in the ice and how they impact the overall strain rate within a floe. ¿ American Geophysical Union 1995