Background: Persistent cue reactivity to drug-related stimuli is a well-known phenomenon among abstinent drug users and has been found to be a predictor of relapse. Cue exposure therapy (CET) aims to reduce this cue reactivity by exposing abstinent drug users to conditioned drug-related stimuli while preventing their habitual response, i.e. drug use. Methods: 127 abstinent heroin-dependent Dutch inpatients were randomized to CET (n = 65; 55 completers) and placebo psychotherapy treatment (PPT) (n = 62; 59 completers). It was examined whether CET would lead to a decrease in drug-related cue reactivity (using mixed-design ANOVA) and subsequently to lower dropout and relapse rates (using logistic regression) compared to PPT. Results: Both groups responded with a similar decrease in self-reported cue reactivity (craving, mood). The CET group did show a significant decrease in physiological reactivity (skin conductance) compared to PPT. However, dropout and relapse rates were, contrary to our expectations, significantly higher in the CET group. Conclusions: This is the first randomized controlled trial showing that CET, compared to a non-specific psychotherapy, might increase dropout and relapse rates among abstinent heroin-dependent clients in a drug-free setting. Caution is warranted when applying CET in this specific context.