Aim: This cross-sectional investigation evaluates the reliability of estimating medial gastrocnemius anatomical cross-sectional area (aCSA) in typically developing and spastic cerebral palsy (SCP) cohorts. It verifies whether muscle volume estimations based on aCSA improve when aCSA is multiplied by muscle–tendon unit (MTU) or muscle length, and whether the resulting errors in volume estimations are smaller than changes after intervention. Method: Fifteen typically developing children (mean age 8y 2mo [SD 1y 5mo], six males, nine females) and 30 children with SCP (mean age 9y 2mo [SD 2y 5mo], 22 males, eight females, Gross Motor Function Classification System [GMFCS] level I=15, II=15) participated in the investigation. The SCP cohort was divided according to GMFCS level. A three-dimensional freehand ultrasound technique was used to estimate medial gastrocnemius aCSA, muscle volume, MTU, and muscle length. Estimated muscle volume (aCSA×MTU or muscle length) was compared with the measured muscle volume. Results: Anatomical cross-sectional area, muscle volume, and muscle length significantly differed between the typically developing and two SCP cohorts (p≤0.050). aCSA multiplied by either MTU or muscle length improved estimations of medial gastrocnemius volume. Leave-one-out cross-validation revealed an absolute difference with measured muscle volume of 3.77 ml (SD 2.90). Interpretation: This investigation revealed that medial gastrocnemius muscle volume can be reliably estimated in a clinically feasible manner in typically developing children and those with SCP. What this paper adds: Medial gastrocnemius anatomical cross-sectional area (aCSA) can be reliably estimated in children with spastic cerebral palsy. The location of the anatomical cross-section should be taken with respect to muscle and not bone length. Medial gastrocnemius volume can be reliably estimated by multiplying aCSA and muscle length. The errors in volume estimations are smaller than reported differences after interventions.