The types, triggers, and spatial distribution of landslides are closely related to the spatial complexity of geological conditions, which are indispensable factors in landslide susceptibility assessment. However, geological conditions often consider only a single index, leading to underutilization of geological information in assessing landslide hazards. In this research, "Geological Complexity" is proposed as a novel index, consisting of tectonic complexity, lithologic complexity, structural complexity and seismicity, representing the spatial heterogeneity of the materials and structures of geological bodies. The contributions of each complexity component to landslides are quantified using a model that combines the weight of evidence and the analytic hierarchy process and then integrated into the Geological Complexity indicator. The performance of this model was further validated with data from the Eastern Himalayan Syntaxis. The results show that the predictive accuracy of Geological Complexity in bedrock-triggered landslide assessment is higher than any individual components and other traditional indicators such as distance to fault and lithology. This suggests that Geological Complexity can more accurately measure the controlling role of geological conditions on the occurrence of landslides than other indices. Selecting multiple geological parameters is more effective than relying on a single parameter.