We examine associations between client attachment style and therapeutic alliance in a 3-arm randomized controlled trial of brief motivational interviewing and cognitive–behavioural therapy compared with longer term motivational interviewing and cognitive–behavioural therapy or standard care alone. Client self-report measures of attachment style were completed at baseline, and both clients and therapists in the treatment arms of the trial completed alliance measures 1 month into therapy. We found that insecure–anxious attachment was positively associated with therapist-rated alliance, whereas clients with insecure–avoidant attachment were more likely to report poorer bond with therapist. There was no evidence that client attachment significantly predicted clinical or substance misuse outcomes either directly or indirectly via alliance. Nor evidence that the length of therapy offered interacted with attachment to predict alliance.