Muscle loss associated with disease (cachexia) or with aging (sarcopenia) is common in dogs, but clinically relevant methods for quantifying muscle loss are needed. We previously validated an ultrasound method of quantifying muscle size in dogs in a single breed. The goal of this study was to assess the variability and reproducibility of the Vertebral Epaxial Muscle Score (VEMS) in other dog breeds. Static ultrasound images were obtained from 38 healthy, neutered dogs of 5 different breeds between 1- and 5-years-old. The maximal transverse right epaxial muscle height and area at the level of the 13th thoracic vertebra (T13) were measured. Length of the 4th thoracic vertebra (T4) was measured from thoracic radiography. Ratios of the muscle height and area to vertebral length (height/T4 and area/T4, respectively) were calculated to account for differences in body size among breeds. Reproducibility testing was performed on 2 dogs of each breed (26% of the total) to determine intra- and inter-investigator reproducibility, as well as intra-class correlation. Mean height/T4 = 1.02 +/- 0.18 and mean area/T4 = 3.32 +/- 1.68. There was no significant difference for height/T4 (P = 0.10) among breeds, but breeds were significantly different in area/T4 (P < 0.001). Intra-class correlation ranged from 0.80 to 0.99. Testing showed better reproducibility for height/T4 compared to area/T4. The VEMS using height/T4 was valid and reproducible for healthy dogs of different sizes and body conformations. Studies assessing this technique in dogs with congestive heart failure and other diseases associated with muscle loss are warranted.