To study the base flow recession at the watershed scale, the log-scale plot of -dQ/dt similar to Q proposed by Brutsaert and Nieber [10] has been used to estimate the recession parameters, i.e., the slope and interception of the theoretical recession slope curve. The lower envelope or the best fit in some studies is usually used to determine the recession slope curve for natural watersheds. However, human interferences exist in most watersheds around the world. This paper discusses the impact of human interferences, which include groundwater pumping, water diversion and return flow, on the determination of the recession slope curve and the cloud shape of the data points of -dQ/dt similar to Q. First, values of -dQ/dt generated for hypothetical watersheds are analyzed. Then real data for three watersheds in Illinois is analyzed to verify the hypothetical analysis. The placement of the recession slope curve depends on the coexistence and relative amount of the evapotranspiration, groundwater pumping or even water diversion if it exists, and the return flow. When the water consumption rate is small, the recession slope curve can even be located at the upper envelope of the cloud of points representing historical data. These results suggest that the use of the lower envelope as a guideline for estimating recession parameters for watersheds subject to human interferences can result in biased estimates. (C) 2010 Elsevier Ltd. All rights reserved.