Muscle ultrasonography is frequently used to improve the understanding of musculoskeletal impairments in children with spastic cerebral palsy (SCP). So far, most studies on muscle morphology and architecture have included typically developing children and children with SCP with similar ancestry, being mainly Caucasian. Less is known about differences in muscle morphology between children with different ancestral backgrounds. Therefore, the aim of this study was to compare muscle morphology and architecture of the medial gastrocnemius in typically developing children with African, South Asian and Southeast Asian descent from Suriname. This explorative cohort study identified children as Maroon (Ghana, African descent), Hindustani (India, South Asian) or Javanese (Indonesia, Southeast Asian), aged 5–10 years. Using 3D freehand ultrasound with the subject prone, the following medial gastrocnemius parameters were defined: muscle tendon unit (MTU) length, muscle belly length, tendon length, muscle volume, muscle thickness, anatomical cross‐sectional area (ACSA), fascicle length, pennation angle, and physiological cross‐sectional area (PCSA). In addition, differences between ancestral groups were assessed for the length of the MTU, muscle, tendon and fascicles in two passive stretch conditions corresponding to an externally applied joint torque of 1Nm and 4Nm. One‐way ANOVA with post hoc t‐tests were used to investigate differences between the ancestral groups. In total, 100 Hindustani (n = 34), Javanese (n = 34) and Maroon (n = 32) children were included. For statistical analyses, we matched the children by age, which resulted in groups of 25 children per ancestral group (n = 75). There were no differences found in MTU length, muscle belly length, ACSA, PCSA and muscle volume. Tendon length, fascicle length and pennation angle were different between ancestral groups. Compared to Javanese children, tendon length was longer (p = 0.001) and pennation angle (p = 0.001) was larger in Maroon children and fascicle length was shorter in both Maroon and Hindustani children (p < 0.001). While there was a difference found in MTU length at different conditions of passive stretch between ancestries, no differences were found in the muscle, tendon and fascicles. This is the first study that investigated macroscopic morphological and architectural parameters for the medial gastrocnemius in one extended cohort of typically developing children, stratified in three ancestral subgroups. The current results imply that ancestry‐specific reference data for children are needed, especially for tendon length, fascicle length and pennation angle when investigating altered muscle morphology in neurological or neuromuscular pathologies, such as SCP. Future studies should report the ancestral background when describing muscle morphology and architecture of children and ancestral specifications should be included in normative databases.