There are several ways to analyze electrochemical noise signals (RE) to observe the activity at the metal - electrolyte interface. Due that records are available at discrete time periods, the most common way to examine the data is through the application of statistical methods in the time domain. These methods can be classified based on their usefulness in those which seek to quantify the magnitude of the corrosion process (standard deviations), highlighting the use of noise and resistance in trying to extract information about the mechanism of the corrosion process itself, based on the assumption that the shape of the temporary registers are a reflection of the behavior of the metal - electrolyte interface. Among these is possible to mention the average value, the bias, the kurtosis, and pitting index, which may reflect changes in the corrosion mechanism. This paper presents the statistical treatment of the data obtained from specimens with RE steel embedded in concrete made ​​with NaCl ( Cl ), NaCl + NaNO2 ( I) without addition of NaCl ( S ) in the mixing water and exposed to an average natural marine .