Researchers confronting their own data often find those data to be more unruly, ill-mannered, and irascible than the well-behaved, cooperative data found in textbook examples. This chapter describes techniques for detecting and taming those nasty data that otherwise could ruin one's analyses. This chapter is devoted to methods for knowing the deviations so that one may more accurately describe nature's ways. For didactic reasons, it considers the detection or diagnosis techniques, and suggests some useful remedies. The methods for detecting nasty and ill-mannered data depend somewhat on the type of analysis. The chapter talks about simple and multiple regressions. It considers several remedial strategies that are likely to be effective. However, the chapter also considers a commonly tried strategy, nonparametric statistics, that is not likely to be either effective or appropriate, while effective remedies such as transformations and removing outliers remain controversial.