This work presents a methodology for evaluating the impact of uncertainties associated with stochastic parameters of equipament on the composite reliability índices. Two methods are used for the estimation of indices variances. The first one makes use of stratified Monte Carlo simulation scheme, where a reliability evaluation is carried out, conditioned to every sample stochastic parameters. In this scheme, besides means and variances, empirical distributions are computed for the reliability indices. The second method employs the sensitives of the reliability indices with respect to variations in the stochastic parameters, obtained as a by-product of the traditional reliability evaluation, to calculate analytically the indices variances. In the stratified Monte Carlo simulation scheme, computed confidence intervals for the variances estimates. Are also computed confidence intervals are used for the evaluating the precision of the results obtained analytically through the sensitivities. The proposed methodology is illustrated in case studies with the IEEE-Reliability Test System and with a configuration of the North-Northeastern Brazilian System, planned for the year 2002.