Onset of psychosis typically occurs in young adulthood, and is likely to be preceded by anxiety, low mood, suspiciousness and perceptual anomalies. Where these develop into mild or brief psychotic experience, and impact on functioning, young people are identified as being at risk of psychosis. Early intervention may reduce severity or delay transition to psychosis, and yield significant healthcare cost savings. This study examined the impact of an online attachment-based imagery task on paranoia, anxiety, mood and self-esteem, in college students. An experimental design was used to compare the effects of secure and anxious-ambivalent attachment imagery. A total of 301 college students, aged 18–48 years (M = 20.1, SD = 2.976), were randomly assigned to one of the two imagery conditions, and assessed pre and post imagery task on standardised measures of paranoia, anxiety, mood and self-esteem. A series of mixed model analyses of variance showed that participants in the secure attachment imagery condition reported lower levels of paranoia, anxiety and negative mood, and higher levels of positive mood and self-esteem, compared with those in the anxious-ambivalent attachment imagery condition. The study is limited by the lack of a neutral control condition and follow-up measures. Nevertheless, study demonstrated the impact of attachment-based imagery on paranoia. If these effects are replicated with ‘at risk mental state’ groups, and maintained at follow-up, online imagery may provide a safe and highly accessible means of attenuating paranoia in young people at risk of developing psychosis.