Life estimation normally requires large size sample of life data. When there is only right-censored life data and the sample size is small, additional assumption or special technique is required. The present paper applies Monte Carlo simulation to generate life samples for a given life distribution, by that a conservative relationship between the population percentile corresponding to 6 and the minimum observation in a sample of size n is characterized. Thereby, (-3) is estimated according to several right-censored data by conservatively taking the censoring time as the minimal observation in a complete life sample of the same size. Such a method is applied to the log-normal distribution of structural part life providing a known variation coefficient (/). The variation coefficient of a particular type of structural part is assigned according to engineering experience.