Climate warming has been and is expected to continue faster in the Arctic than at lower latitudes, which generates major challenges for adaptation. Among others, long-term planning of development of socio-economic infrastructure requires climate-based forecasts of the frequency and magnitude of extreme flood events. To estimate the cost of facilities and operational risks, a probabilistic form of long-term forecasting is preferable. A stochastic model allowing to simulate the probability density function (PDF) of hydrological variables based on a projected climatology, without modelling hydrological time series, is applied to estimate extreme flood events caused by spring snow melting in the Russian Arctic. The model is validated by cross-comparison of modelled and empirical PDFs using historical time series. The PDF parameters of spring flood runoff are assessed in a regional scale under the SRES and RCP climate scenarios for 2010–2039. For the Russian Arctic, an increase of 17–23 \% in the mean values and a decrease of 5–16 \% in the coefficients of variation of the spring flood runoff are expected. Territories are outlined where engineering calculations of the extreme maximum discharges should be corrected to account for the expected climate change. The extreme maximum discharge for a bridge construction over the Nadym River is calculated.