Abstract: The natural river flow in the Haihe River basin has been continuously reducing in recent years, which leads to significant changes particularly in the wet and dry conditions of hydrology. As such, research on the calculation method of wet and dry probability catering for nonstationary hydrological series is needed. Based on the standardized runoff index and GAMLSS model, a new method computing expected number of occurrences (ENO) and expected waiting time (EWT) was proposed according to different levels of wet and dry hydrological events. Then, we further investigate on the evolution law of probability of wet and dry of natural river flow in Haihe River basin under changing environment. Our main finding are outlined as follows: ① The probability of wet and dry of surface runoff showed a significant trend of increasing low flow and decreasing high flow. ② Compared with other probability distributions including the traditional stationary hypothesis, the LOGNO distribution with time(t) as the covariate has the best performance when fitting the surface runoff series. The ENO of wet and dry hydrological events in the historical period calculated based on this distribution is closer to the actual situation. ③ The EWT of extremely dry and wet hydrological events in the Haihe River basin calculated under different scenarios of the nonstationary optimal model were 4.9—9.4 a and 14.5—36.0 a respectively, indicating that the considerably higher likelihood of extreme dry event in oncoming future.