In some countries, the electric power systems are operating near to their limits due to the absence of investments in the transmission network expansion and natural growth of the electricity demand. This operation condition can also occur in electric power systems in which the transmission expansion is carried out in appropriate way. In this case, the excessive loading of the transmission network is usually originated by the loss of interconnections that transport large energy blocks. The two operation scenarios described above have caused Voltage Stability problems in the electric power systems. The voltage instability states are mainly characterized by the presence of two mechanisms: the unsolvability of the power flow equations and the controllability loss. The disturbances that originate these two mechanisms are of stochastic nature. Consequently, the voltage instability indices, used to analyze the unsolvability and controllability loss, are random variables. In this way, the voltage stability assessment would recognize the uncertainties associated with the parameters of the electric network, for example: load fluctuations and equipment availability. Generally, the uncertainties modeling in the voltage stability is carried out using the following probabilistic methods: the Monte Carlo Simulation and the State Enumeration. The main index estimated by these methods is the voltage instability risk. However, the voltage instability risk evaluation is usually carried out considering only one of the mechanisms that cause voltage instability scenarios. Furthermore, the severity of the unstable states has not been properly investigated. The aim of this thesis is to develop a method to carry out a probabilistic assessment of the voltage stability that take into account the two mechanisms that cause the voltage instability in the evaluation of its risk. Probabilistic indices, based on Well-Being Analysis, are also proposed to express the severity of the voltage instability states. The proposed method is based on the combination of the following techniques: State Enumeration Method, Monte Carlo Simulation, D’ Matrix Method and Nonlinear Optimal Power Flow. The State Enumeration and Monte Carlo Simulation Methods are used to select the system states resulting of equipment failures and load forecast errors. The identification of the controllability loss and the solvability restoration of the power flow equations for the selected states are carried out by the D’ Matrix Method and by the Nonlinear Optimal Power Flow, respectively. The combination of the methods cited above was used to obtain the following probabilistic indices: voltage instability risk, expected value of the voltage instability margin for the buses, and Well-Being states probabilities. The results of the tests with the proposed method revealed that the probabilities of unstable states, associated with the two voltage instability mechanism, are very significant. Additionally, the Well-Being Analysis was able to identify the root cause and the severity of the voltage instability problems.