Bioluminescence imaging (BLI) of gene expression in live animals is a powerful method for monitoring development, tumor growth, infections, healing, and other progressive, long-term biological processes. BLI remains an effective approach for reducing the number of animals needed to monitor dynamic changes in gene activity because images can be captured repeatedly from the same animals. When examining these ongoing changes, it is sometimes necessary to remove rhythmic effects on the bioluminescence signal caused by the circadian clock’s daily modulation of gene expression. Furthermore, BLI using freely moving animals remains limited because the standard procedures can alter normal behaviors. Another obstacle with conventional BLI of animals is that luciferin, the firefly luciferase substrate, is usually injected into mice that are then imaged while anesthetized. Unfortunately, the luciferase signal declines rapidly during imaging as luciferin is cleared from the body. Alternatively, mice are imaged after they are surgically implanted with a pump or connected to a tether to deliver luciferin, but stressors such as this surgery and anesthesia can alter physiology, behavior, and the actual gene expression being imaged. Consequently, we developed a strategy that minimizes animal exposure to stressors before and during sustained BLI of freely moving unanesthetized mice. This technique was effective when monitoring expression of the Per1 gene that serves in the circadian clock timing mechanism and was previously shown to produce circadian bioluminescence rhythms in live mice. We used hairless albino mice expressing luciferase that were allowed to drink luciferin and engage in normal behaviors during imaging with cooled electron-multiplying-CCD cameras. Computer-aided image selection was developed to measure signal intensity of individual mice each time they were in the same posture, thereby providing comparable measurements over long intervals. This imaging procedure, performed primarily during the animal’s night, is compatible with entrainment of the mouse circadian timing system to the light cycle while allowing sampling at multi-day intervals to monitor long-term changes. When the circadian expression of a gene is known, this approach provides an effective alternative to imaging immobile anesthetized animals and can removing noise caused by circadian oscillations and body movements that can degrade data collected during long-term imaging studies.