Secure Message Transmission (SMT) is a two-party protocol by which the sender can privately transmit a message to the receiver through multiple channels. An adversary can corrupt a subset of channels and makes eavesdropping and tampering over the corrupted channels. Fujita et al. (GameSec 2018) introduced a game-theoretic security notion of SMT, and showed protocols that are secure even if an adversary corrupts all but one of the channels, which is impossible in the standard cryptographic setting. In this work, we study a game-theoretic setting in which all the channels are corrupted by two or more independent adversaries. Specifically, we assume that there are several adversaries who exclusively corrupt subsets of the channels, and prefer to violate the security of SMT with being undetected. Additionally, we assume that each adversary prefers other adversaries’ tampering to be detected. We show that secure SMT protocols can be constructed even if all the channels are corrupted by such rational adversaries. We also study the situation in which both malicious and rational adversaries exist.