Grain boundary engineering through the control of grain boundary character distribution (GBCD) has been extensively employed as a powerful tool for achieving enhanced properties and for development of high performance both structural and functional polycrystalline materials. Many efforts were made firstly to increase the frequency of low-energy CSL boundaries of polycrystalline materials in grain boundary engineering. However, the connectivity of grain boundaries can be an important microstructural parameter governing bulk properties of polycrystalline materials as well as the GBCD. In the present work, the connectivity of random grain boundaries was quantitatively evaluated using both the triple junction distribution and random boundary cluster length on the basis of SEM-EBSD/OIM observations, and then these evaluated parameters were linked to intergranular corrosion of SUS304 stainless steel. We have found that the length of the maximum random boundary cluster drastically decrease with increasing CSL boundaries in the fraction ranging 60 – 80% CSL boundaries, which leads to percolation threshold occurring at approximately 70±5% CSL boundary fraction (at 30±5% random boundary fraction). The experimentally observed percolation threshold is much higher than theoretically obtained one based on randomly assembled network (at 35% resistant bonds for a 2D hexagonal lattice). In addition, the fraction of resistant triple junctions is found to increase with increasing the the CSL boundary fraction. An increase in the frequency of resistant triple junctions can enhance intergranular corrosion resistance of polycrystalline austenitic stainless steel even if the GBCD is the same.