Objectives The present paper proposed and tested a methodology for reducing individual's threat response to compassion‐imagery, by increasing their levels of state attachment‐security. Design A total of 68 University students (63% female, mean age = 25) completed an experimental study, where they were randomly assigned to either a 10‐min attachment‐prime (to enhance attachment‐security) or an interpersonal skills module (control condition). Methods Participants completed a compassion‐focused imagery exercise before and after the manipulation, to determine the effects of the attachment‐prime. To measure the effects of the manipulation on individual's threat response, heart rate variability data were collected at baseline and during both compassion exercises. Results As predicted, individuals who reported higher levels of anxious and avoidant attachment styles were more likely to display a threat response (decreases in heart rate variability), to the first compassion‐focused imagery. After receiving an attachment‐prime, heart rate variability increased suggesting that individual's experienced greater self‐soothing responses and decreased threat responses to the second compassion‐focused imagery. Conclusions The present findings suggest that individuals with insecure attachments are likely to require additional support increasing their attachment‐security, before they can successfully engage in compassion‐based exercises or therapies. Practitioner points Compassion‐based exercises may result in fear and consequently avoidance in some populations of individuals. Threat responses to compassion can be reduced by using attachment‐based techniques. Research findings will help inform and broaden the clinical applicability of compassion‐based therapies.