A scale-invariance analysis of rainfall retrieved during the Tropical Ocean Global Atmosphere Coupled Ocean-Atmosphere Response Experiment (TOGA-COARE) campaign is discussed. As already found in the previous Global Atmospheric Research Program (GARP) Atlantic Tropical Experiment (GATE) rainfall data set, these new analyses of high-intensity storms confirm the evidence of scale invariance under self-similar space-time transformations. A simple interpretation of this space-time self-similarity accounting for the hierarchical organization of precipitation patterns is proposed. Finally, a downscaling model based on a log-Poisson generator is calibrated on the results of the multifractal analysis and applied to the generation of synthetic fields, reproducing observed statistical properties over a wide range of space scales and timescales.