Aiming at the uncertainty of Threat Level Assessment (TLA) data sources under modern complex battlefielda fuzzy Bayesian network TLA method was proposed by integrating the fuzzy set theory into Bayesian networks.After well considering about the effect of the distance and azimuth angle of threat sources relative to a UCAV on its stealth capabilitythe threat level was evaluated by integrating such uncertain factors as weatherthreat typedistance and azimuth angle based on randomization of uncercain knowledge.Then a TLA Bayesian network was established by adopting Netica software of the Norsys Software Company in Canadaand simulation was carried out.The simulation results show that the method can assess the threat level rapidly and accurately.