Abstract: Precipitation-runoff relationship is a core issue in engineering hydrology. Copula functions have advantages and flexibility in capturing dependences between variables. This study adopted an Archimedean Copula function to construct such a distribution relationship and detect its abrupt variation points based on the monthly and yearly scales of precipitation and runoff. The Jinghe river basin features precipitation and runoff following the Gamma and logarithmic normal distributions respectively, and according to Copula function analysis its abrupt point on precipitation-runoff relationship occurred in 1996. At the abrupt point, we divided the time series into two parts for the periods 1960~1996 and 1997~2010. Then, the analysis shows that a significant change in the relationship occurred, i.e. a reduction of 30% in water yield during 1997~2010 relative to 1960~1996 in the same conditions of precipitation. And the less precipitation, the greater reduction in water yield, with its maximum of 37.59% at P=500mm. This abrupt variation point on precipitation-runoff relationship calculated by Copula function is demonstrated by considering the variation in human activities, etc.