The sampling variance of a T-year flood when estimated using a curve-f
itting method results from the errors in hydrologic observations. Plot
ting positions, and model-fitting. This paper develops a method to qua
ntify the contribution of plotting positions to the sampling variance
of the T-year flood magnitude. Application of the method to 150 flood-
flow data sets of 41 rivers in the People's Republic of China show tha
t the errors due to plotting positions contribute more to the sampling
variance than others.