The use of electrochemical noise (ECN) as a method to characterize the corrosion-protection properties of organic coatings on metal substrates was pioneered by Skerry and Eden, and since then has been used by others as a probe for coating metal corrosion studies. However, no statistical examination of the reproducibility of the data from such measurements has been published. In the data we present, we have done a systematic analysis of important experimental variables in such systems. We have examined the method for accuracy and reproducibility with respect to sample preparation, sample immersion, and metal substrate preparation. We have taken several marine coatings systems typical of U.S. Navy use, prepared duplicate samples of coating metal systems, and examined them under the same immersion exposure. The variables we considered for reproducibility are paint application (in three-coat systems), metal panel preparation (grit-blasted steel), and immersion conditions. We present ECN data with respect to immersion time on the values of noise voltage standard deviation σV, noise current standard deviation σl, and the noise resistance Rn as given by σV/σl. The variation among supposedly identical sample pairs in identical immersion monitored under identical conditions is presented. The statistics of the time records of the data are considered, and the variations with respect to specific coatings classes are also considered within the limits of the data. Based on these data, comments concerning ECN on coated metal systems as a predictive test method are presented along with special considerations that must be made to properly use the method for coating ranking and lifetime prediction.