Archer, Graham-Kevan, and Davies (2005) provide a number of criticisms of a 2001 meta-analysis that examined the relationship between testosterone and aggression, and discuss the criteria for study inclusion and the coding of study characteristics. This paper addresses the issues raised by Archer et al. (2005). It discusses the exclusion criteria and choices made by Archer et al. (2005) in their analysis, and provides a reanalysis of the data in light of the concerns that Archer et al. (2005) raised. In the reanalysis, the results support the original claim that there is a small, positive relationship between testosterone and aggression. The relationship is moderated by age (in males) and time of day, as in the original analysis. For the most part, Archer et al. (2005) bring a very important point to the forefront. It is important to be careful and precise in conducting meta-analyses given the ease with which errors in coding occur. Tables, references