Abstract This paper presents a probabilistic low cycle fatigue (LCF) assessment for geometric tolerances on high load contact surfaces of gas turbine compressor blades. The typical patterns of the geometric deviations for the root contact flank of the compressor blades are identified and characterized according to coordinate measurement machine (CMM) measurements of the blade root. These typical patterns are closely related to the root form manufacture tools and process. Finite element (FE) models for the typical geometric deviation patterns are created based on nodal coordinates transformation of the surface nodes on the blade root contact flanks and radius. An optimized blade root profile tolerance is established, which enables significant cost saving. The elastic-plastic FE analysis with nonlinear contact model, material strain-life test, response surface, and constrained Monte Carlo simulation are used to create a probabilistic LCF life model for the optimized tolerance. The model quantifies the effect of the geometric deviations, blade mass, and material property on the blade LCF life. The result shows that with the optimized tolerance, the probability of blade LCF failure is very low and acceptable. It is also shown that the strain life material property is the most critical factor for the LCF failure. The root profile tolerance and blade mass are seen to have a much weaker effect on the blade LCF life.