In this study, we investigated the accuracy of using off-line bioluminescence imaging (BLI) and tomography (BLT) to guide irradiation of small soft tissue targets on a small animal radiation research platform (SARRP) with on-board cone beam CT (CBCT) capability. A small glass bulb containing BL cells was implanted as a BL source in the abdomen of 11 mouse carcasses. Bioluminescence imaging and tomography were acquired for each carcass. Six carcasses were setup visually without immobilization and 5 were restrained in position with tape. All carcasses were setup in treatment position on the SARRP where the centroid position of the bulb on CBCT was taken as “truth”. In the 2D visual setup, the carcass was setup by aligning the point of brightest luminescence with the vertical beam axis. In the CBCT assisted setup, the pose of the carcass on CBCT was aligned with that on the 2D BL image for setup. For both 2D setup methods, the offset of the bulb centroid on CBCT from the vertical beam axis was measured. In the BLT-CBCT fusion method, the 3D torso on BLT and CBCT was registered and the 3D offset of the respective source centroids was calculated. The setup results were independent of the carcass being immobilized or not due to the onset of rigor mortis. The 2D offset of the perceived BL source position from the CBCT bulb position was 2.3 mm ± 1.3 mm. The 3D offset between BLT and CBCT was 1.5 mm ± 0.9 mm. Given the rigidity of the carcasses, the setup results represent the best that can be achieved with off-line 2D BLI and 3D BLT. The setup uncertainty would require the use of undesirably large margin of 4–5 mm. The results compel the implementation of on-board BLT capability on the SARRP to eliminate setup error and to improve BLT accuracy.