Microarray technology, while it is no longer in its infancy, is still immature and unpredictable. If it works well, microarray expression analysis is a powerful method for estimating the behavior of 10,000 genes simultaneously. But how often does it work well? Because of the rapid adoption of this young technology, we have to anticipate that one of the consequences will be the occasional publication of incorrect results. Some microarray data are simply wrong. In the course of trying to perfect microarray gene expression experiments in the last few years, we think that we have encountered most of the problems that can lead to erroneous interpretation of microarray data. The goal of this chapter is to offer advice from our experience in the hope that it will increase the odds of obtaining accurate data from microarray experiments in neurobiology.