Pre-pubertal boys and girls speak with acoustically different voices despite the absence of a clear anatomical dimorphism in the vocal apparatus, suggesting that a strong component of the expression of gender through the voice is behavioural. Initial evidence for this hypothesis was found in a previous study showing that they can alter their voice to sound like a boy or like a girl. However, whether children can spontaneously modulate these voice components within their own gender in order to vary the expression of their masculinity and femininity remained to be investigated. Here, seventy-two English-speaking children aged 6 to 10 were asked to give voice to child characters varying in masculine and feminine stereo-typicality to investigate whether primary school children spontaneously adjust their voices sex-related cues in the voice – fundamental frequency (F0) and formant spacing (ΔF) – along gender stereotypical lines. Boys and girls masculinised their voice, by lowering F0 and ΔF, when impersonating stereotypically masculine child characters of the same sex. Girls and older boys also feminised their voice, by raising their F0 and ΔF, when impersonating stereotypically feminine same-sex child characters. These findings reveal that children have some knowledge of the sexually dimorphic acoustic cues underlying the expression of gender, and are capable of controlling them to modulate gender-related attributes, paving the way for the use of the voice as an implicit, objective measure of the development of gender stereotypes and behaviour.