Child temperament is a predictor of non-verbal ability (i.e., thinking and problem-solving skills that do not fundamentally require verbal language production and comprehension). Given that temperament scores might vary depending on whether the reporter is a parent or a teacher, this study analyzes (a) whether those reports are different and (b) how each report predicts child non-verbal ability in a non-western, educated, industrialised, rich and democratic (non-WEIRD) sample. The Matrix subtest of KABC-II (a non-verbal ability task widely used in non-WEIRD contexts) was administered to 85 Argentinian children (47 girls, 38 boys) aged 4-5 years, from middle-to-low socioeconomic status homes. Also, the Child Behaviour Questionnaire-Very Short Form (CBQ-VSF) was administered to obtain temperament reports from parents and teachers. Factors from parents' and teachers' reports did not correlate. Only factors including items from the effortful control dimension from teachers' reports predicted Matrix total score with a small effect size (partial [eta-squared] = 0.10). Instead, the parents' report did not predict non-verbal ability. Our results provide infrequent data from non-WEIRD low-SES populations and underscore the unique and predictive validity of teachers' perspectives in clarifying the connections between child temperament and nonverbal ability.