Predicting the life and reliability of boiling water nuclear reactor (BWR) structural components subject to stress corrosion cracking requires high quality data and data analysis for both crack initiation and crack propagation. Laboratory study and prediction of crack propagation rates is more advanced relative to crack initiation studies. This is partially attributed to the probabilistic nature of crack initiation experimentation which requires many test specimens to be statically valid. This paper briefly reviews concepts used in applying reliability methods and data analysis in developing quality data and then applying these concepts to IGSCC crack initiation data. Since the distribution of time to crack
initiation tends to be highly skewed, these time to failure data are not well represented by measures of central tendency such as the mean time to failure. Data from creviced and sensitized 304SS are better described by either a lognormal or Weibull distribution. For the Weibull distribution... Keywords: sample size, Weibull distribution, lognormal distribution, crack initiation, IGSCC