To demonstrate the reliability requirements of aeronautical systems, as prescribed by the ARP4761A standard, and which may present complex failure scenarios, fault tree analysis is recommended. These qualitative and quantitative approaches are mainly based, for the estimation, on modelling assumptions of exponential distributions. However, for mechanical systems, these assumptions are not always respected. Moreover, for safety failure events, the ARP4761A prescribes extremely low failure probabilities per flight hour, on the order of 10⁻⁹ to 10⁻⁵. In structural reliability, FORM-System can handle multiple failures scenarios and very low level of failure probability. It is still used as a reference method for recent research developments in 2025 [1]. Monte Carlo-System simulations are another method. However, the total number of sampling must check 10^(k+2) to 10^(k+3) to demonstrate a P_(f_SYS)=10^(-k). For about ten years, Active Learning Kriging-Based Reliability Methods have been the scientific references for the uncertainty propagation [2]. Therefore, AK-SYS [3], or more recent AK-SYS-IE [4], are efficient solutions to estimate failure probabilities of systems. In the continuation, a novel method, Active Learning Kriging-Based – Importance Sampling for System Reliability allows to model multiple failure scenarios and demonstrate, through stochastic modelling and simulation, the achievement of extremely low failure probability targets. Furthermore, to guarantee a high level of systems reliability, it is necessary to demonstrate adequate conformity rates for manufacturing processes. AK-IS-SYS is applied both to estimate these probabilities of systems failures and the non-conformity rates of associated processes. AK-IS-SYS is an incremental approach, based on AK-IS proposed by Echard & al. [5] and AK-SYS by Fauriat & al. [4]. However, original failure probability or non-conformity rate estimators, based on the IS estimators, are defined to integrate the failure domains for series systems. The AK-IS-SYS performances are validated by the accuracy of its low probability estimations and the required number of high fidelity deterministic structural or process simulations on industrial examples.