Attenuation due to propagation is a significant problem for weather radars operating in C-band or at higher frequencies. When an accurate rainfall rate estimate is required, as in hydrological applications, a good attenuation correction technique must be carefully applied. The task becomes more complex when only incoherent radar observables are available (absolute and differential reflectivity Z H and Z DR): in such a case, standard cumulative algorithms for attenuation correction are commonly used on a range cell by range cell basis. However, instabilities show up in the presence of anomalous reflectivity data and can diverge, leading to unrealistic rainfall rates. To avoid the correction become useless or even void of sense, it must be applied on data that have been previously corrected from other sources of errors. The major causes of anomalous reflectivity are related to beam blocking effects and related clutter phenomena, common in regions with a complex orography. Here we analyze on experimental data the advantages obtained by accounting for clutter and screening effects, before correcting attenuation due to propagation. Results show that the preliminary correction of radar data can help to reduce instabilities and to obtain better rainfall rate estimates when compared to raingage data. Some residual instability may show up in some cases, due to the inadequacy of the reflectivity-attenuation relationships at high reflectivity values.