Since 1984, the Urban Community of Lyon (CO.UR.LY.) has been setting up a network of rain gauges with a density of about 1 station per 20 km. This network is integrated in the French system of Experimental and Representative Basins. To get a better knowledge of the spatial distribution of the rainfall, a study was based on the establishment of I.D.F. relationships for each of these measuring stations. Since the study period extended only over 5 to 6 years, only return periods of 1 and 2 years were taken into account. The results show a high statistical heterogeneity of rain gauging data. These discrepancies may be due to sampling errors, to systematic measuring errors, to an excessively short observation period or to real spatial differences in rainfall. Each one of the causes of error was studied, and these 4 points all seem to play a more or less significant role in the differences observed between the stations. Thus there may be a spatial heterogeneity of the rainfall over the area of the CO.UR.LY. If confirmed, this observation may have very important consequences and may question risk assessment in urban hydrology.