BACKGROUND: We used bioluminescence imaging (BLI) to validate serial assessment of neoplastic growth within the murine liver. We hypothesized that, in mice bearing luciferase-expressing liver metastases, bioluminescence would reflect neoplastic burden estimated by liver weight. MATERIALS AND METHODS: Murine colon carcinoma cells were infected with a retroviral vector encoding luciferase. Bioluminescence was measured in vitro, and in vivo following intrasplenic tumor cell inoculation into Balb/C mice. At varying time intervals, mice were imaged and immediately killed to determine correlation of in vivo and ex vivo BL with liver weight. To examine sensitivity in vivo, we performed direct intrahepatic inoculation. RESULTS: In vitro, photon emission correlated with increasing cell numbers. In the metastatic model, a statistically significant increase was depicted in both liver weight and bioluminescence 20 days after tumor inoculation compared with earlier time points (P < 0.0001). With progressively increasing tumor burden, however, a poor correlation between in vivo BL and liver weight was observed. When livers with very large tumors were excluded, R(2) = 0.90. In contrast, correlation between ex vivo BL and liver weight remained high throughout the study period (R(2) = 0.94). CONCLUSIONS: When imaging tumors at end-stages of disease, in vivo BL underestimates actual tumor burden. If context-specific limitations are recognized, however, BLI is a sensitive approach to temporal monitoring of tumor progression in preclinical models.