Methods that are designed to be resistant to even multiple severe outliers. Such methods are an extreme example of robust statistics.