Evidence suggests that attachment styles may influence subclinical psychosis phenotypes (schizotypy) and affective disorders and may play a part in the association between psychosis and childhood adversity. However, the role of attachment in the initial stages of psychosis remains poorly understood. Our main aim was to describe and compare attachment styles in 60 individuals at ultra high risk for psychosis (UHR) and a matched sample of 60 healthy volunteers (HV). The HV had lower anxious and avoidant attachment scores than the UHR individuals (p < .001). Sixty-nine percentage of the UHR group had more than one DSM-IV diagnosis, mainly affective and anxiety disorders. The UHR group experienced more trauma (p < .001) and more mood and anxiety symptoms (p < .001). Interestingly, in our UHR group, only schizotypy paranoia was correlated with insecure attachment. In the HV group, depression, anxiety, schizotypy paranoia, and social anxiety were correlated with insecure attachment. This difference and some discrepancies with previous studies involving UHR suggest that individuals at UHR may compose a heterogeneous group; some experience significant mood and/or anxiety symptoms that may not be explained by specific attachment styles. Nonetheless, measuring attachment in UHR individuals could help maximize therapeutic relationships to enhance recovery.