<p>This study addresses the event-triggered asynchronous fault detection problem for continuous-time semi-Markovian jump systems subject to communication delays. First, an adaptive memory event-triggered transmission mechanism is proposed by using historical data to adjust the triggering threshold dynamically, which makes the strategy more flexible and meanwhile reduces unnecessary data transmission. Second, mode asynchronism between the system and FD observer is considered and modeled via a hidden Markov process since the complex environment and system’s stochastic characteristics. Third, an improved delay-dependent Lyapunov function is constructed for the system analysis and synthesis, relaxing positive definite limitations and thus yielding the reduced conservatism. Based on this, a sufficient criterion is derived to co-design the asynchronous observer gains and triggering parameters. A practical simulation demonstrates the effectiveness of the developed method, with the results showing both fast detection after fault occurrence and a reduction in data transmission amount.</p>