Behavioural differences among animals of the same species are often attributed to differences in temperament. Animals with different temperaments may have different welfare states as they differ in their ability to adapt to environmental challenges. Some species have been classified as either active or passive copers. Previous methods for measuring temperament have been criticized as they were considered subjective and did not measure consistency over time. The current study attempted to address these concerns when measuring the temperament of horses. Three tests were used to record the behaviour of 33 horses: an arena test, response to a person and response to an object. The tests were repeated three times in order to determine if the results were consistent. The horses were consistent in their responses during the arena test. The horses responded similarly to the person and object test, but did not show consistency in their responses over time. Responses during the arena test were not found to predict responses during the person or object test. The horses could not be classified as either active or passive based on the results of these behavioural tests. The responses during the behavioural tests could not be used to predict responses during a startle test or during reintroduction to the group after testing. Their responses were not correlated with their status in the group or with behavioural ratings given by the farm team leader. Due to the high variability in the results of the response to person and object tests, the only test that could be used to reliably assess temperament in horses was the arena test.