Emotion-based decision making (EBDM) is the capacity to make decisions based on prior emotional consequences of actions. Several neuropsychological tasks, using different gambling paradigms and with different levels of complexity, have been designed to assess EBDM. The Bangor Gambling Task (BGT) was created as a brief and simple card gambling-task to assess EBDM. BGT contains a single-card deck and requires participants to decide whether to gamble or not, which can result in wins or losses. Unknown to the participant, the winning probabilities decrease throughout the task (from 0.75 in the first block to 0.25 in the fifth block), requiring participants to reduce their gambling probability to avoid long-term losses. A few studies have offered evidence regarding the BGT convergent validity. However, there are no computerized versions of BGT available, thus slowing the process of gathering information to explore the EBDM mechanisms behind the task, its validity, and clinical usefulness. In this article, we present a computerized version of the BGT using the Matlab environment and make all our code available. We explore BGT's replicability and analyze its probabilistic structure, providing trial-level and block-level analyses. Eighty-one participants performed the computerized version, which followed the same structure as the original version. It took participants 8.5 ± 3.3 minutes to complete the task, which is faster than the paper version. Replicating previous studies, participants diminished their gambling probability throughout the task, learning to inhibit the initially rewarded gambling behavior. This change in gambling probability could be considered a proxy for EBDM. Our analyses suggest that the last blocks are especially sensitive to capturing deficits in EBDM, and we propose some modifications to BGT's original version to enhance the initial exploratory and learning phase. Our results show that the BGT constitutes a quick and simple task to evaluate EBDM capacities.