In order to estimate how strongly nodes affect each other in a supply chain network, prior studies have primarily relied on the eventual operation statuses of suppliers recorded at the end of each risk propagation process, and thus learn only the structure of influence relationships (i.e., the existence of influence relationships) rather than the strengths of influence relationships. However, compared with the structure of influence relationships, their strengths provide greater utility for accurately predicting which suppliers are likely to be affected when a supply chain risk occurs. In this paper, we investigate the problem of how to learn the strengths of influence relationships between nodes in a supply chain network. To this end, we construct a likelihood function with the strengths of influence relationships as variables, and then solve the likelihood function to obtain the strengths of influence relationships that are most likely to generate eventual operation status data. Results from large-scale experiments across real and simulated networks provide strong evidence of our approach’s reliability and computational efficiency.