Abstract: To synthesize recession coefficients in the Xinanjiang model, this paper presents a relationship of the coefficients versus the time interval and storage coefficient for linear reservoirs based on the linear reservoir theory, and discusses potential uncertainties due to longer time intervals and input variability within the intervals. For the recession coefficient of channel water, which is very difficult to calibrate in daily rainfall-runoff analysis, a new approach is adopted to replace its direct calibration by synthesizing the storage coefficients of linear reservoirs. Results show that the new approach is very successful in the estimation of the hourly recession coefficients of 13 small basins of areas in the range of 100 to 3000 square kilometers and has achieved their values with an average error of 0.005 and a standard deviation of 0.022. This approach allows a new way to synthesize the recession coefficients in the Xinanjiang model based on vast previous studies about river concentration.