Abstract Eating behaviours have been associated both with being underweight or overweight and poor growth. The Children's Eating Behaviour Questionnaire (CEBQ) is a widely used measure of child eating behaviours. The instrument is, however, mostly validated in high-income countries, with a scarcity of evidence among developing countries such as Ethiopia. The present study aims to assess the cultural adaptability and validity of the CEBQ to be used in Ethiopia. We conducted a school-based cross-sectional study among 542 caregivers of children aged 3–6 years in selected preschools. Tests of factorial validity, convergent validity and reliability were performed. The Confirmatory Factor Analysis model indicated that eight subscales provided the best fit (root-mean-square error of approximation = 0⋅05 (90 % CI 0⋅045, 0⋅055); Comparative Fit Index = 0⋅92 and Tucker–Lewis Index = 0⋅90) after seven items from the original CEBQ were removed. Convergent validity with child's weight status was found for emotional overeating, food fussiness, satiety responsiveness and slowness in eating subscales. Reliability, measured using Cronbach's α, provided values between 0⋅50 and 0⋅79. The eight-factor structure of the CEBQ showed adequate content validity and provided factorial, discriminant and convergent validity among preschool children. Further replication of the study among low-income countries is essential to improve the literature on children's eating behaviours.