Artificial selection was used to establish different levels of agonistic behavior in laboratory-reared wild female Mus musculus. A within-family selection design with replicated high and low lines and two nonselected control lines was employed. Females only were tested at 8 weeks of age on 2 consecutive days. Testing consisted of placing a C57BL/6 female mouse in the home cage of the isolation-reared wild mouse for 7 min or until an attack occurred. The wild females were rated on a 5-point scale for agonistic behavior, and the sum of the scores over the 2 test days was the criterion for selection. The six lines, each containing ten breeding pairs, were selected for four generations. By the fourth selected generation, the responses of the high and low lines had diverged in the expected directions. One-way analyses of variance indicated reliable differences among high, low, and control lines in the second, third, and fourth generations. It was concluded that female agonistic behavior is influenced by genotype and that the level of this behavior can be manipulated by means of artificial selection.