Floods are natural hazards which are very common in Slovenia. Protective measures require accurate and reliable quantification of the designed discharge which we can get using frequency analysis. Flood frequency analysis is a procedure which connects major flood occurrence with its frequency. When there are various samples of the same data available regional frequency analysis can be used. It is based on regions combining different water gauging stations with similar properties. During the analysis not only data for each station but the information for the whole region is used. The main advantage of the regional flood frequency analysis is that estimated values are more accurate especially for stations with lack of data. There are four main steps to perform regional flood frequency analysis. At first, accuracy and discordance of data has to be checked. The second and also the most important step is identification of homogeneous regions. Many grouping methods can be used but the most practical method of forming regions for the large data sets is cluster analysis. At the same time, regional homogeneity can also be checked using a heterogeneity measure. The last two steps are the choice of an appropriate frequency distribution for a region, based on goodness-of-fit tests and estimation of the parameters and quantiles of the selected distribution. In the first part of the thesis theoretical background of statistical methods used is described and in the second part regional flood frequency analyses for Slovenia were performed.