Traditional methods used in hydrologic engineering to estimate extreme river discharges of different return periods are based on the assumption of stationarity in the probability distribution function of the series of extreme flows. This hypothesis collapses owing to the effects of climate change, climate variability, land use/land change, and deforestation on the hydrological dynamics, which are reflected into the series of hydrologic variables. We illustrate such effects on the hydrology of Colombia, and implement two models for estimating extreme river discharges using mixed probability distribution functions, which take into account the effects of climate variability on the series of extreme discharges. Application of these methods for Colombian rivers allowed us to conclude that the Mixed Type I Weibull distribution provides an adequate estimation for low flows, and the Fréchét Mixed Type I distribution for peak flows and floods.