In this paper, the development of a 3-D failure criterion for saline ice is presented. The need for such general 3-D failure formulation stems from the fact that, during ice-ship interactions, ice undergoes a complex state of deformation and stress before it fails and breaks away, and the use of the uniaxial strength of ice to compute impact ice loads may lead to inaccurate load calculations and non-conclusive results. In recent years, with the availability of High Power Computers (HPC), numerical methods are being used more than ever before in marine and ice engineering problems. Numerical models based on computational techniques such as finite elements, boundary elements and discrete elements require 3-D constitutive models and failure criteria to represent the behavior of the materials involved (such as the behavior of the ship structure, ice, and water "fluid"). At high-speed mpacts (strain rates >10-3 s-1), ice behaves as a linear elastic material with a brittle mode of failure. Previously, Derradfi-Aouat [Derradji-Aouat, A., 2000. A unified failure envelope for isotropic freshwater ice and iceberg ice.ASME/OMAE-2000, Int. Conference on Offshore Mechanics and Arctic Engineering, Polar and Arctic section New Orleans, US, PDF file # OMAW-2000-P?A # 1002] developed a unified 3-D failure envelope for both fresh water isotropic ice and iceberg ice. In this paper, that formulation is extended to include failure of saline ice (in addition to fresth water ice and iceberg ice.) The results of a significant number of true triaxial tests using Laboratory Grown Ice (LGSI) were obtained from the open literature. The results of these tests formed a database that enables the existing failure model [Derradji-Aouat, A., 2000. A unified failure envelope for isotropic freshwater ice and iceberg ice. ASME/OMAE-2000, Int. Conference on Offshore Mechancis and Arctic Engineering, Polar and Arctic section, New Orleans, US, PDF file # OMAE-2000-P/A # 1002] to be extended from the isotropic fresh water ice and iceberg ice to columnar saline ice. Mroz's [J. Mech. Phys. Solids 15 (1967) 163] concept for the multi-surface failure theory is used in both studies (the present study, for saline ice, as well as in the previous study, for the fresh water isotropic ice and iceberg ice). It appears that the same set of the equations is applicable to the failure of all three types of ice. The possibility of the existence of a universal and general failure criterion for all types of ice is discussed. The validation of the present multi-surface failure criterion was discussed on the basis of comparisions between predicted failure curves and actual true triaxial test results. An overall discrepancy of predicted versus measured strength values of less than 20% was calculated.