AbstractData obtained from four national samples of 6–11 year old children are presented, showing preference for four countries, perceived friendship of these countries with the subject's own country and perceived relationships among the target‐countries. Two quantitative analogues of Heider's balance theory are used to examine the relations between these sets of data, one employing a multiplicative and the other a distance approach. Both models were found to fit the data adequately, further research being needed to discriminate between them. An explanation in terms of ‘ethnocentric correspondence’ is given to explain why prediction is noticeably better for British and Austrian than for Belgian and Greek