Abstract Objective: Heavy alcohol use and alcohol dependence are major public health issues, and in-person treatment has been impaired by the restrictions necessary during the Covid-19 pandemic. This thesis follows the development of a smartphone-based cognitive bias modification training, which heavy drinkers and patients treated for alcohol dependence can access for at-home use. Method: Several versions of a smartphone-based alcohol approach avoidance training (AAT) were developed and tested through three pilot experiments (n = 26, n = 25, n = 13). The newly developed AAT web-app was then validated in two short online studies on heavy drinkers with one-day (n = 62) and one-week follow-ups (n = 107). Finally, the web-app was used for a three-months randomized control clinical trial in patients leaving clinic treatment for alcohol dependence (n = 598). Results: Heavy drinkers showed a reduction in craving scores and problem alcohol use at follow-up after only 20 – 30 minutes of training on the AAT web-app. Alcohol dependent patients exhibited high adherence to the web-app, accessing it for over 50 training sessions on average. Feedback for the web-app was positive, and suggestions for improvement were collected and addressed through updates of the web-app. While an intention to treat analysis showed no significant differences between training and control groups, secondary subgroup analyses of participants who accessed the app for regular training point to efficacy of the active training when used for at least five to fifteen sessions. Participants who used the active version of the AAT web-app for at least five sessions showed reduced craving. An analysis of preliminary data points to lower relapse rates at one-year follow-up in patients who used the active training web-app for at least fifteen sessions. Conclusions: The web-app was developed and piloted from a basic prototype to a three-months intervention with gamification elements. Effects on craving and problem alcohol use were assessed through several studies with concurrent measures. A preliminary analysis with relapse data points to an effect of active training on abstinence in alcohol-dependent participants who accessed the app for at least 15 training sessions. The web-app was well liked, and participants showed high adherence. This new web-app is now in use as part of standard care in an in-patient treatment centre for alcohol dependence in Germany.