[en] Precipitation is one of the most important variables needed in different hydrological models: infiltration, soil loss, droughts, overland flow production, floods, etc. To predict its behavior is complex due to it is highly intermittent over time. Because of the adequate time-frequency representation of wavelet techniques, they are being widely applied to different hydrological resources applications. In this work, wavelet analysis has been applied in order to forecast monthly precipitation data in Mediterranean Coast (Málaga, Southern Spain) using Artificial Neural Networks (ANN) models. Several mother wavelets have been evaluated at different decomposition levels for rainfall predictions using a standard multilayer perceptron architecture. The results obtained indicate that the Daubechies wavelet transforms of order 5 (db5) used at level 3 are the most appropriate for this case study, deriving the more effective performance of all the models assessed.