The volitional illuminance preferences of 9 closely related, captive-born cactus mice were determined by allowing them a choice between 10 illuminance levels ranging from darkness to 1.1 lux. The levels could be altered bidirectionally to brighter or dimmer values, one step at a time in predetermined sequence, by instrumental means. The mice were confined in exercise-wheel enclosures and tested and retested at ages of 4 to 35 months. For all 9 animals, the time spent running in the wheel had marked, minor, incipient, or presumptive peaks at level 5 (0.0014 lux) or level 6 (0.0025 lux) ; these peaks probably reveal the optimum illuminance range for the animals' vision. But the major peak in the active time of most individuals was in darkness (or extremely dim light). There was generally a marked tendency for the animals to spend a greater percentage of inactive time than active time in darkness. Five animals retested after a lapse of at least 5 months showed changes in illuminance preferences, there being a tendency for the minor preference for dim light either to increase with age or to appear first long after maturity. In addition, one 5-month-old mouse showed a remarkable series of changes in illuminance preferences over a period of several weeks. The illuminance regimes to which small nocturnal mammals are exposed are analysed in terms of a paradigm involving two sets of factors-illuminance preferences and factors unrelated to these preferences. It is proposed that, although both sets of factors determine illuminance exposures in the wild, only factors in the illuminance preference category determine illuminance exposures (selectivity) in captivity. The major preference of cactus mice for darkness and the lahility of their preferences are analysed in terms of this paradigm.