In the analysis of high-throughput data, a very common goal is the detection of genes or of differential expression between two groups or classes. A recent finding from the scientific literature in prostate cancer demonstrates that by searching for a different pattern of differential expression, new candidate oncogenes might be found. In this chapter, we discuss the statistical problem, termed oncogene outlier detection, and discuss a variety of proposals to this problem. A statistical model in the multiclass situation is described; links with multiple testing concepts are established. Some new nonparametric procedures are described and compared to existing methods using simulation studies.