Abstract: REASONS FOR PERFORMING STUDY: Behavioural tests as well as observers' ratings have been used to study horses' temperament. However, the relationship between the ratings and the responses in behavioural tests has not yet been studied in detail. OBJECTIVES: The aim of the present study was to examine this relationship between ratings and responses. METHODS: Eighteen mature Swedish Warmblood horses were subjected to 2 behavioural tests, one relating to novelty (novel object test) and one to handling (handling test). Subsequently, 16 of these horses were ridden by 16 equally experienced students, having no former experience with the horses. Immediately after each ride, the students scored the horse for 10 temperamental traits using a line rating method. RESULTS: It was shown that for each temperamental trait all 16 riders agreed on the ranking of the horses (0.212<W<0.505, P < 0.001). CONCLUSIONS: Correlations between behavioural and heart rate variables in the behavioural tests revealed that horses with a high level of locomotion or much restlessness behaviour exhibited high mean heart rate and low heart rate variability. In particular, heart rate variables in the behavioural tests were found to correlate with riders' rating scores. Furthermore, the underlying components of the handling test, retrieved with a principal component analysis (PCA) correlated with riders' rating scores while the underlying components of the novel object test did not. POTENTIAL CLINICAL RELEVANCE: It is concluded that it is possible for a large panel of assessors to agree upon a horse's temperament and that objective measures from behavioural tests correlate significantly with temperamental traits assessed by a panel of assessors.