Linear elastic fracture mechanics (LEFM) is of limited use in analyzing fracture tests conducted on laboratory-sized specimens of sea ice or saline ice. To extrapolate the results obtained from laboratory experiments to larger sizes, nonlinear fracture mechanics must be invoked. The use of linear elastic fracture mechanics in the case of laboratory-sized specimens is invalidated—in general—by active near crack tip deformation mechanisms such as microcracking, grain boundary sliding, and plastic slip. The fictitious crack model, which makes use of the stress-separation curve, can be used to incorporate this process zone into the fracture analysis so that fracture results from laboratory-sized specimens can be meaningfully interpreted. The stress-separation curve for any type of ice has not yet been determined. In this paper, stress-separation curves for saline ice are constructed such that the response computed using the fictitious crack model matches the experimental results. Crack plane orientations both parallel and perpendicular to the direction of the growth of the columnar grains are considered. Two different stress-separation curves are needed to model the crack growth in these two orientations. The fictitious crack model is implemented using the weight function method. Various physical parameters that affect the obtained stress-separation curve are discussed.