Mental imagery manipulations are used to treat several psychological disorders, but their utility in treating cocaine use disorder (CUD) is unknown. Using prompted re-experiences and simulations with contrasting valence, we assessed the acute impact of a deliberate mental imagery task on cocaine craving. Design A quantitative-qualitative 'mixedmethods' analysis of data collected for a randomized controlled trial that was stopped prematurely. Setting UK National Health Service addictions treatment clinic and outpatient clinical research facility (laboratory). Participants Adults with CUD. The original target sample was 120. All participants enrolled at the point the original trial was stopped were included (38 enrolled, 31 completed study). Interventions Personalized (3-minute) cue-exposure (handling cocaine paraphernalia and watching video of drug preparation), immediately followed by a single 5-minute, audio-recorded, self-guided and verbally described imagery task with random assignment to one of four conditions: two mental imagery memory re-experiences (positive image before initiation to cocaine use or a negative image of a 'worst time' adverse cocaine use episode) or two future simulations (positive theme of recovery from CUD or negative theme of worsened CUD).Measurements Task transcripts were rated for imagery detail using five dimensions using a six-point scale of imagery detail (ID) (total score = 0-25) and thematically coded. The outcome measure was cocaine craving using the Craving Experiences Questionnaire-strengths version (CEQ-S11; score = 0-110) reported at baseline, arrival at the laboratory, and immediately after the cue-exposure and mental imagery tasks. Findings A mixed-effects, longitudinal, restricted linear regression, with the past-positive imagery condition as referent, showed main effects of reduced craving after the imagery task (b = -29.2, 95% confidence interval (CI) = -45.3 to À13.1, P-value < 0.001) and increased craving for the futurenegative task (b = 14.2, 95% CI = 0.1-28.4, P-value 0.049). There was a future-negative task by post-imagery craving interaction (b = 28.1, 95% CI = 0.1-56.1, P-value 0.049). A theory-driven, deductive/inductive qualitative analysis of the transcripts revealed six major themes: sensory characteristics, CUD vicious cycle, self-care, emotions and appraisals, social role and CUD recovery. Positively themed simulations included interpersonal connections and rewarding activity; negative images included personal adversity, with appraisals of self-criticism and hopelessness. Transcripts with more imagery detail were not associated with significantly greater reductions in craving in the positive or negative imagery task (r = -0.32, 95% CI = -0.69 to 0.13 and r = 0.06, 95% CI = -0.58 to 0.53, respectively). Conclusion In people with cocaine use disorder, after cue-exposure, a self-guided imagery task with positive themes reduced craving, whereas mental imagery simulating worsened cocaine use disorder did not appear to.