Background: The Undergraduate Medical and Health Sciences Admissions Test (UMAT) is used to select medical students in Australia and New Zealand but empirical evidence of its construct validity has never been reported. Aims: To identify the underlying constructs assessed in each of the three sections of the UMAT. Based on conclusions from an early qualitative study (Mercer & Chiavaroli 2006), it was expected that Section 1 scores would correlate with scores obtained from standard measures of cognitive ability (verbal and numeric reasoning), Section 2 scores would correlate with emotional intelligence, and Section 3 scores would be most strongly related to abstract or non-verbal reasoning ability. Method: Final year high school students (n=432) completed tests of numerical, verbal, and non-verbal cognitive ability, and emotional intelligence. Correlations and multiple regressions assessed the relationship of these tests with scores on each section of the UMAT. Results: UMAT Section 1 was significantly related to verbal, non-verbal, and numerical reasoning tests. Section 2 was significantly related to emotional intelligence and verbal reasoning, but the majority of variance in this section's scores remained unexplained. Section 3 scores significantly correlated with non-verbal and numerical reasoning. Conclusions: The UMAT Sections 1 and 3 appear to be tests of cognitive abilities. Further research is required to identify the constructs being measured by Section 2.