We have identified a generalized arousal component in the behavior of mice. Analyzed by mathematical/statistical approaches across experiments, investigators, and mouse populations, it accounts for about 1/3 of the variance in arousal-related measures. Knockout of the gene coding for the classical estrogen receptor (ER-alpha), a ligand-activated transcription factor, greatly reduced arousal responses. In contrast, disrupting the gene for a likely gene duplication product, ER-beta, did not have these effects. A combination of mathematical and genetic approaches to arousal in an experimentally tractable mammal opens up analysis of a CNS function of considerable theoretical and practical significance.