The major aim of my thesis project has been to develop a non-human primate model of trait anxiety, using a new world monkey, the common marmoset. The first step was to identify animals high or low in trait anxiety. Based on the findings that high trait-anxious individuals display over-generalization of fear responses, a pathogenic marker of elevated trait anxiety in humans, a new aversive discriminative conditioning paradigm was designed. Testing a normal cohort of marmosets revealed that 26% of the animals displayed both behavioural and physiological signs of fear generalization, i.e. failure to discriminate safety from danger cues (‘failed’ group). The remaining 74% showed successful discrimination (‘passed’ group). Additional regression analysis on several behavioural and physiological responses early in training revealed two potential biomarkers of high trait anxiety in marmosets: suppressed baseline blood pressure, indicative of contextual effects, and hyper cue-specific vigilance. These measures predicted the animal’s likelihood of passing or failing the discrimination. The finding that the ‘failed’ group showed intact discriminative performance in the appetitive domain rules out an interpretation of the results in terms of a general impairment in learning, per se. To further determine whether these hypothetically high trait-anxious animals would display enhanced anxiety-related responses in more classical primate models of anxiety, human intruder and rubber snake tests were performed on a large sample of marmosets. Principal component analysis on multiple behavioural measures revealed two components underlying performance: ‘emotionality’ and ‘coping strategy’. Although no difference was found in the human intruder test, the ’failed’ group displayed significantly elevated levels of ‘emotionality’ in comparison to the ‘passed’ group in the rubber snake test. Moreover, the two biomarkers of fear over-generalisation also reliably predicted the ‘emotionality’ scores. Finally, having developed a marmoset model of trait anxiety, investigations into the neural underpinnings, especially prefrontal involvement in trait anxiety mechanisms, were carried out by testing the animals on two cognitive flexibility tests: an orbitofrontal cortex (OFC)-dependent incongruent object discrimination test and a lateral prefrontal cortex (lPFC)-dependent detour reaching rule transfer test. Whilst group differences did not reach significance, the two biomarkers of fear over-generalisation, the suppressed baseline blood pressure and hyper cue-specific vigilance, were inversely and differentially correlated with perseverative performance on the two tests, the lPFC- and OFC-dependent tests, respectively. This not only indicates that high trait anxiety can lead to improvements in certain aspects of prefrontal cognitive function but also suggests that changes in the activity of at least two distinct prefronto-subcortical neural circuits, a cue-sensitive amygdala-OFC and a context-sensitive hippocampus-lPFC circuit, may contribute to trait anxiety.