Children’s sex-typed preferences for colours and toys are well-established, and often function as markers of sex-typicality in research on the development of sex-typed behaviour. However, children’s sex-typed colour and toy preferences have not been tested cross-culturally, or in remote unindustrialised cultural settings. The present thesis tested children’s preferences for sex-typed toys in four cultural settings: Shipibo villages in the Lake Imiria region of the Peruvian Amazon; kastom villages in the mountains of Tanna Island in Vanuatu in the South Pacific; children attending school in Lenakel town on Tanna Island; and in a large industrialised city in Australia. It also tested children’s colour preferences in three of these cultures. It was hypothesised that colour and toy preferences would show some similarities across cultures, and further, that similarities in toy preferences across cultures would be explained by the different types of play afforded by the toys. Results suggested that colour preferences, specifically, a sex difference in preference for pink, are specific to industrialised cultures. Results further suggested that some sex differences in toy preferences replicate in different cultures, and that the relationship between toy preferences and children’s preferences for play affordances is a potentially important area for further research. The present thesis also provided two demonstrations of how new statistical methods, adapted from complex and dynamic systems theory, could be applied to the cross-cultural dataset. A machine learning method suggested that sex, more than culture, affects children’s sex-typed toy preferences. A multistate dynamic method further suggested that sex, more than culture, affects the dynamics of children’s toy choices.