Current mammalian bioreporters using either firefly luciferase (luc) or GFP constructs require lysis and/or exogenous excitation to evoke a measurable response. Consequently, these cells cannot serve as continuous, on-line monitoring devices for in vivo imaging. Bacterial luciferase, lux, produces a photonic reaction that is cyclic, resulting in autonomous signal generation without the requirement for exogenous substrates or external activation. Therefore, lux-based bioluminescent bioreporters are the only truly autonomous light-generating sensors in existence. Unfortunately, the bacterial lux system has not yet been efficiently expressed in mammalian cells. In this research, three approaches for optimal expression of the a and b subunits of the bacterial luciferase protein were compared and reporter signal stability was evaluated from stably transfected human embryonic kidney cells. Maximum light levels were obtained from cells expressing the luciferase subunits linked with an internal ribosomal entry site (IRES). Cells harboring this construct produced bioluminescence equaling 2.6 X 106 photons/sec compared to 7.2 X 104 photons/sec obtained from cells expressing the luciferase from a dual promoter vector and 3.5 X 104 photons/sec from a Lux fusion protein. Furthermore, the bioluminescence levels remained stable for more than forty cell passages (5 months) in the absence of antibiotic selection. After this time, bioluminescence signals dropped at a rate of approximately 5% per cell passage. These data indicate that mammalian cell lines can be engineered to efficiently express the bacterial lux system, thus lending themselves to possible long-term continuous monitoring or imaging applications in vivo.