Precise assessment of fracture-healing is vital in both clinical and research settings, where therapies are typically assessed using animal models with union as the study end point. Radiographic scoring systems have been developed for clinical use; however have not been validated in pre-clinical models. Thirty sets of radiographs of rat tibial shaft fractures, treated with external fixation were reviewed by six observers using the Radiographic Union in Tibia (RUST) scale, the Lane & Sandhu score, and an overall impression of union. Fleiss's kappa and Intra-class Correlation Coefficients (ICC) were used to determine reliability. Inter-observer and intra-observer agreement using the general impression score were moderate [kappa; 0.58; 95%CI (0.49-0.65) and 0.66 (0.43-0.89), respectively]. Inter-observer and intra-observer agreement were excellent using both the RUST score [ICC; 0.81 (0.72-0.89) and 0.86 (0.74-0.93), respectively], and Lane & Sandhu score [ICC; 0.88 (0.81-0.93) and 0.90 (0.81-0.95), respectively]. Employing a defined scoring system enhances both the reproducibility and repeatability of bone healing assessment in a small animal model. Routine reporting of fracture scoring methodology should be encouraged to enrich results and facilitate data synthesis across studies. © 2014 Orthopaedic Research Society. Published by Wiley Periodicals, Inc. J Orthop Res.