Some of the fundamental principles of the Quantitative Precipitation Forecasts from operational numerical atmospheric models are shortly addressed in the paper. The attention is then focused on the relations among the QPF uncertainty, the subgrid variability of the forecasted precipitation patterns and the uncertainty of the hydrologic prediction. Recognizing the fundamental deterministic nature of numerical model outputs, possible probabilistic parametrizations for the forecasted precipitation are analyzed with the use of a conceptual non linear hydrologic model. The definition of the scores which commonly define the quality of the QPF products is also reviewed in the same probabilistic framework. Some research topics to be developed in the field of QPF related to flood prediction are finally suggested.