The aim of this work is to compute the stabilization energy E stab (n) of ͓X(H 2 O) n ͔ Ϫ (XϵF, Br, and I for nϭ1 -60) clusters from Monte Carlo simulations using first-principles ab initio potentials. Stabilization energy of ͓X(H 2 O) n ͔ Ϫ clusters is defined as the difference between the vertical photodeachment energy of the cluster and the electron affinity of the isolated halide. On one hand, a study about the relation between cluster structure and the E stab (n) value, as well as the dependence of the latter with temperature is performed, on the other hand, a test on the reliability of our recently developed first-principles halide ion-water interaction potentials is carried out. Two different approximations were applied: ͑1͒ the Koopmans' theorem and ͑2͒ calculation of the difference between the interaction energy of ͓X(H 2 O) n ͔ Ϫ and ͓X(H 2 O) n ͔ clusters using the same ab initio interaction potentials. The developed methodology allows for using the same interaction potentials in the case of the ionic and neutral clusters with the proviso that the charge of the halide anion was switched off in the latter. That is, no specific parametrization of the interaction potentials to fit the magnitude under study was done. The good agreement between our predicted E stab (n) and experimental data allows us to validate the first-principles interaction potentials developed elsewhere and used in this study, and supports the fact that this magnitude is mainly determined by electrostatic factors, which can be described by our interaction potentials. No relation between the value of E stab (n) and the structure of clusters has been found. The diversity of E stab (n) values found for different clusters with similar interaction energy indicates the need for statistical information to properly estimate the stabilization energy of the halide anions. The effect of temperature in the prediction of the E stab (n) is not significant as long as it was high enough to avoid cluster trapping into local equilibrium configurations which guarantees an appropriate sampling of the configurational space. Parallel tempering method was applied in particular cases to guarantee satisfactory sampling of clusters at low temperature.