Background: Muscle mass influences performance and health in horses. It is important to be able to easily evaluate muscle mass in horses. Objectives: To develop an equine muscle condition score (MCS) to be used by veterinarians, horse owners, trainers and other people involved in horse management and training. Study design: Observational, cross-sectional study. Methods: Twenty-five Thoroughbred horses at different training levels were included. Initial assessment included body condition score, height and bodyweight. Seven muscles (m. splenius, m. brachiocephalicus, m. multifidus, m. longissimus dorsi, m. gluteus medius, m. semitendinosus and m. tensor fascia latae) were imaged using ultrasonography and measured. Repeatability of ultrasonographic measurements was tested in advance by measuring each muscle 10 times in two horses. Muscle sizes were compared between two groups; trained and untrained to determine difference in muscle mass. Horses were subsequently ranked according to measured muscle mass, and palpable and visual differences between horses at different ranks were identified. A 5-point MCS was developed. The MCS was tested by an independent, blinded operator in all 25 horses. Results: Ultrasonographic measurement of all muscles had a coefficient of variation under 5%. All muscle sizes were significantly correlated to each other except for m. brachiocephalicus, which was not correlated to m. semitendinosus and m. multifidus. All muscles were significantly larger in trained horses than in untrained horses except for m. brachiocephalicus and m. splenius. After testing, minor adjustments were made to optimise usability of the score. Main limitations: The study population consisted of very similar horses. Further studies are needed to test the reliability of the MCS in horses of different breeds, disciplines, body condition scores and health. Conclusion: This study resulted in an equine MCS that could identify differences in muscle mass between horses. The resulting MCS was similar to other muscle assessment systems developed independently from our study, thus strengthening the findings.