Streamflow prediction in ungauged basins is considered as a real challenge worldwide. Considerable efforts have been devoted to tackle this issue. However, no single method can be considered as the best solution and employed everywhere. So, the main objective sought for this work is to provide continuous monthly streamflow information in ungauged catchments in northern Algeria, needed for planning and water resources management. Thus, the first part of this work is devoted to the evaluation of the performance of a monthly water balance‑based rainfall-runoff model, namely GR2M, in gauged catchments in northern Algeria. The rainfall-runoff modeling step allowed the delineation of the zones, within the study area, where the GR2M model is suitable. In order to delineate homogeneous regions, prior to regionalization of the GR2M model’s parameters, two classification techniques have been assessed (principal component analysis (PCA) and self-organization maps (SOM)). Regional models between the model’s parameters and features of the catchments have been derived. The rainfall-runoff modeling results show that the water balance‑based GR2M model is not recommended for semi-arid regions. On the other hand, the regionalization results show that the regionalized models, based on the SOM classification technique, led to better performances than those obtained by PCA. For most of the catchments, considered as ungauged, the monthly and annual streamflow values estimated using the SOM-regionalised models are very promising.