To address the threats of radiation effects on the reliability in space environments，the radiation probabilistic risk assessment （PRA） methodologies for multi-core parallel fault-tolerant radiation-hardened computer systems （CCUs） are investigated.First，the single-event upset （SEU） rate distribution models for the critical memory devices，e.g.，programmable read only memory （PROM），magnetoresistive random access memory （MRAM），and static random access memory （SRAM），are developed based on the ground-based simulation test data.Three candidate empirical distributions，i.e.，exponential distribution，Weibull distribution，and log-normal distribution，are analyzed.The optimal fitting function is selected according to the maximum entropy principle，and it is shown that the SEU rates of the memory devices follow the log-normal distribution most reasonably.Second，a device failure probability model is established under the framework of the error detection and correction （EDAC） mechanisms.A system-level reliability modeling method is constructed for the CCUs，and the Monte Carlo simulation is adopted to derive the failure probability distribution over a 5-year mission lifecycle.Finally，the probability importance，structural importance，and relative importance of the components are analyzed，and the critical risk sources are quantified，by which the device hardening strategies and risk mitigation recommendations are suggested.The findings provide a theoretical foundation and quantitative support for the design of highly reliable radiation-tolerant systems.