This paper proposes a fuzzy methodology for dam failure risk assessment, having developed a case study with the Simplicio hydroelectric power plant located in the Center Western region of Brazil. The fuzzy inference was obtained by calculating the Critically Index in a first step, and then the defuzzified Risk Priority Number in a second step. Through a FMECA there were introduced estimates of failures in the plant related to the overtopping, mass movement of the gravity concrete dam (global instability), excessive deformation and blocking access of the crest, backrest instability caused by material movement, piping through the dam, water contamination, rapid lowering of the reservoir and foundation failure. The most relevant issue with a moderate magnitude detected by the traditional FMECA approach is the piping erosion (concealed internal erosion) caused by the excessive seepage due to insufficient or inadequate foundation, which may due to the improper design or construction. The fuzzy expert system, on the other hand, classified the risk magnitude as critical, caused by the high severity of the damage that might be caused by the piping erosion. The impact might cause several casualties, catastrophic and irreversible damage to the environment and a massive economic loss exceeding $100M. Moreover, the detectability is low, because the operation is not likely to detect the cause. Such result has practical consequences, because increasing the technical concern regarding the foundation stability would result in a more frequent inspection schedule and an enhancement in the detection capability under the consideration of the As-Low-As-Reasonably-Practicable (ALARP) operational procedure.