Floods, being the most prevalent natural disasters, result in significant economic losses and have a profound impact on people’s lives as well as infrastructures. Flood susceptibility mapping can help enhance mitigating strategies. The primary objective of this study is to delimit the flood-prone area within the high Oum Er Rbia watershed. To achieve this objective, an extensive initial field survey was conducted, coupled with the interpretation of Google Earth images. These efforts allowed us to identify a total of 60 flood locations in the area under study. Out of these historical flood locations, 70% selected for training the models, 30% were reserved for the validation process. By partitioning the historical flood locations into training and validation datasets, we aim to develop robust models that can accurately delineate flood-prone areas. This approach ensures that the models are tested against independent data, thereby enhancing their reliability and applicability. Fifteen determining factors including, elevation, slope, aspect, curvature, Stream Position Index, Topographic Position Index, Topographic Wetness Index, Topographic Ruggedness Index, distance from the road, distance from the river, stream density, lithology, Rainfall, Normalized Difference Vegetation Index, land use/land cover, were employed to map the flood susceptibility in the study region. The relationship between each class of flood factors and frequency of floods was determined using the weighting factor, weight of evidence, and frequency ratio approaches. The natural break approach was used to classify all of the factors into nine classes after they were resamples with a pixel size of 30 by 30. Finally, a hazard map was produced and categorized into five susceptibility levels: very low, moderate, high and very high. The area under the curve (AUC) of the receiver operating characteristic (ROC) was used to validate and assess the performance of models. The outcomes demonstrate that the Weight of Evidence on the training dataset had a high performance (Woe = 76%, FR = 71%, WF = 73%), and the performance of models, of 71%, 73%, and 75% respectively. The outcomes of this study could be very helpful in managing flood risk and can be applied by agencies that deal with flood risk management, planners, and local disaster management.