Metal containers represent a situation where a specific metal is exposed to a wide variety of electrolytes of varying degrees of corrosivity. For example, hundreds, if not thousands of different products are packaged in an aluminum beverage can. These products vary in pH, chloride concentration and other natural or artificial ingredients which can effect the type and severity of potential corrosion. Both localized (perforation) and uniform corrosion (metal dissolution without the onset of pitting) may occur in the can. A quick test or series of tests which could predict the propensity towards both types of corrosion would be useful to the manufacturer. Electrochemical noise data is used to detect the onset and continuation of pitting corrosion. Specific noise parameters such as the noise resistance (the potential noise divided by the current noise) have been used to both detect pitting corrosion and also to estimate the pitting severity. The utility of noise resistance and other electrochemical parameters has been explored through the application of artificial neural networks. The versatility of artificial neural networks is further demonstrated by combing electrochemical data with electrolyte properties such as pH and chloride concentration to predict both the severity of both localized and uniform corrosion.