Alexithymia is a personality trait characterised by difficulties in identifying and describing one’s emotions, constricted imaginal processing, and an externally oriented cognitive style. Alexithymia is associated with psychopathology and interpersonal problems. The aim of the current study was to evaluate the psychometric properties of the most frequently used measure of alexithymia, the self-report 20-item Toronto Alexithymia Scale (TAS-20). Specifically, the study aimed to (1) cross-validate the hypothesised three-factor structure of the TAS-20, (2) determine whether the measure indirectly assesses the constricted imaginal thinking component of alexithymia, despite its absence of imagination items, (3) examine the overlap between the TAS-20 and measures of psychopathology, and (4) determine whether the TAS-20 assesses actual, rather than merely perceived, emotional understanding. Participants were 194 (138 female) university students and community members who completed an online survey. Confirmatory factor analyses showed mixed support for the hypothesised three-factor model of the TAS-20; however, this model provided a better fit to the data than either a one- or two-factor model. Inverse relationships were found between the TAS-20 and measures of perspective-taking and fantasy (although this relationship was only marginally significant for fantasy). There were moderate to large positive associations between the TAS-20 and measures of depression, anxiety, stress, and negative affectivity. Inverse relationships were found between the TAS-20 and objective measures of emotional ability; however, these relationships were no longer significant after the effects of negative emotions and affectivity were partialled out. Higher TAS-20 scores were also associated with more moderate affective valence ratings of emotion-evoking stimuli, and thus lower self-reported arousal. Together, these findings suggest problems with the TAS-20’s construct validity. Theoretical and practical implications are discussed.