The spatial scaling properties of annual average streamflow is examined usi
ng records from 1 433 river basins across the continental United States. Th
e log-linear relationship In(E[Q(i)(r)]) = a(r) + b(r)In(A(i)) is represent
ative throughout the United States, where E[Q(i)(r)] represents the expecta
tion of the rth moment of annual streamflow at site i, and A(i) represents
drainage area. The scaling model parameters a(r) and b(r) follow nearly per
fect linear relationships a(r) = r alpha and b(r) = r beta throughout the c
ontinental United States. We conclude that the probability distribution of
annual streamflow follows simple scaling relationships in all regions of th
e United States. In temperate regions where climate is relatively homogeneo
us, scale alone describes most of the variability in the moments of annual
streamflow. In the more climatically heterogeneous regions, such as in the
Upper Colorado and Missouri river basins, scale alone is a poor predictor o
f the moments of annual flow.