The variation of temporal continuous rainfall properties with spatial scale was investigated by analyses of daily 11.2-year rainfall time series of point values and spatial averages obtained from a network of 161 rain gauges in southern Sweden. The number of series studied ranged from 16 for point values to 1 for 8000 km2, the latter corresponding to the total network area. The analyses included investigations of general descriptive statistics, power spectra, empirical probability distribution functions, and scaling of statistical moments. Two important characteristics of the rainfall process in the region were indicated. The first was that temporal statistical properties related to extreme values qualitatively changed at a spatial scale of ~2800 km2, whereas properties of the mean process did not. Although this change may partly be related to changes in the sample size, its presence does raise some important questions concerning the transformation between point values and regional scales in hydroclimatological rainfall modeling. The second important characteristic was a temporal scaling behavior between 1 day and 1 month with properties depending on the spatial averaging area. This finding is in agreement with previous investigations where a spatial variation of temporal scaling properties has been found and suggests the possibility of scaling-based time series modeling where the regional parameters are adjusted to take local climatological factors affecting the rainfall process into account. ¿ 1999 American Geophysical Union