This paper studies the evolution of hydrological regimes of Alpine rivers. The detection of past changes is a key factor to understand the issues of water management in the Alps. This presentation shows the main results of a trend analysis at local and regional scales. A new dataset including more than 118 daily discharge time series was collected in the six countries of the Alpine space: France, Germany, Switzerland, Slovenia, Austria, and Italy. These time series contain at least forty years of daily records, with six series over one hundred years. This dataset covers the whole spectrum of hydrological regimes existing in the Alps, from pure glacier- and snowmelt-regimes to mixed rainfall / snowmelt regimes. Gauging stations were selected to represent natural regimes, i.e. from undisturbed catchments. This selection was made using metadata collected from data producer in each country, in order to discard the catchments with major anthropogenic influences such as discharge control structure, water supply or intake. Most studied rivers have snowmelt or ice-melt driven regimes. The former regime is characterized by one low flow seasons during winter (precipitations are stored in the catchment as snow). Both regimes have a major high flow season in spring (summer for glacier regimes) corresponding to the thaw of the snow accumulated during the winter. A set of hydrologic indices were defined to characterize the hydrologic regimes in terms of low, medium and high flows. Concerning low flows, the threshold level method was used to define drought events (Fleig 2006). The low flow threshold was taken as the 15% quantile of the flow duration curve. Seasonality indices were computed to quantify the time of start, centre and end of the drought. Volume deficit, duration, and minimum discharge values for each event were also calculated. Medium flows were described by the annual mean flow and the base flow index. For high flows, a particular interest was paid to snowmelt-related flows. The snowmelt component was taken as the baseflow, according to the Talaksen and Van Lanen (2004) baseflow separation method. Snowmelt spring floods could thus be identified for each river, and were studied regarding their seasonality and intensity. The seasonality was described by the time of start, centre and end of the flood (Stewart et al. 2005). Intensity was characterized by the baseflow maximum discharge and the cumulated baseflow volume. There are many ways to detect changes in long time series of hydrological data. In this study, trend detection was implemented in two steps. In a first step, at-site statistical tests were applied for each hydrologic indice to all gauging stations. Different statistical tests can be used to evaluate the stationarity of time series, depending on the proprieties of the data. Non-parametric tests were favoured because no a priori distributional assumption could be made for most indices. We therefore used the rank-based Mann-Kendall test, modified to take into account the possible auto-correlation existing in the data (Hamed et Rao 1998). In a second step, the consistency of at-site results was tested using a regional procedure proposed by Renard et al. (2008). This procedure evaluates the existence of a trend common to all stations within a homogeneous region, after transformation of at-site series using a normal-score transformation. The results concerning winter droughts have mixed consistency. No consistent trends are found regarding their seasonality, except for the drought end which happens earlier within the season (23% of the studied stations have a significant downward trend with a type-I error level α = 10%). The results suggest a small tendency to less severe winter droughts : Annual minimum is globally increasing (28% of significant upward trends), and annual volume deficit and drought duration are decreasing (26% and 27% of significant downward trends). Results for the spring snowmelt floods show more consistency. A large number of stations present changes in the thaw seasonality: flood start and flood centre occur earlier in the season (51% and 31% of significant downward trends), whereas the flood end seems to happen later, leading to a longer flood duration (53% of significant upward trends). Results concerning the flood intensity and volume are more contrasted. At this time the regional analysis has not been finalized yet, the results will be detailed during the oral presentation. Observational evidences of changes in hydrological regimes are difficult to highlight (Svensson et al. 2006), especially for extremes of the hydrological regime. This paper studies a number of hydrological indices describing droughts and snowmelt-related high flows. A few changes are found regarding low flows in the Alps, with the droughts tending to be less severe. Clearer changes are found concerning the timing of spring snowmelt-related flows ; precocity of start and centre of snowmelt flows, increased duration.