Ultrasonography is widely used to measure gastrocnemius muscle architecture; however, it is unclear if values obtained from digitised images are sensitive enough to track architectural responses to clinical interventions. The purpose of this study was to explore the reliability and determine the minimal detectable change (MDC) of gastrocnemius medialis (GM) and gastrocnemius lateralis (GL) muscle architecture using ultrasound in a clinical setting. A trained sonographer obtained three B-mode images from each of the GM and GL muscles in 87 volunteers (44 males, 43 females; 22±9 years of age) on two separate occasions. Three independent investigators received training, then digitised the images to determine intra-rater, inter-rater, and test-retest reliability for fascicle length (FL), pennation angle (θ) and muscle thickness. Median FL, θ, and muscle thickness for GM and GL were 53.6-55.7 mm and 65.8-69.3 mm, 18.7-19.5° and 11.9-12.5°, and 12.8-13.2 mm and 15.9-16.9 mm, respectively. Intra- and inter-rater reliability of manual digitisation was excellent for all parameters. Test-retest reliability was moderate to excellent with intraclass correlation coefficient (ICC) values ≥0.80 for FL, ≥0.61 for θ, and ≥0.81 for muscle thickness, in both GM and GL. The respective MDC for GM and GL FL, θ, and muscle thickness was ≤12.1 mm and ≤18.00 mm, ≤6.4° and ≤4.2°, and ≤3.2 mm and ≤3.1 mm. Although reliable, the relatively large MDC suggest that clinically derived ultrasound measurements of muscle architecture in GM and GL are more likely to be useful to detect differences between populations than to detect changes in muscle architecture following interventions.