In hydrological analyses and flood risk studies, the application of probability laws and probability distribution functions is inevitable. Hence, in this study at the Karoon watershed, the regional method that has more accuracy compared to other methods along with the linear moment method was used. In this method, homogeneous units over the watershed area are firstly determined; and for each unit an appropriate distribution function is then sought. Using daily rainfall data of 38 rain gauge stations, 3 homogenous units were indentified in the study area. For each unit, an appropriate distribution function was determined using the linear moment method. The distribution functions of Gen. Logistic, Gen. Extreme Value and Gen. Logistic were introduced for homogenous units 1, 2 and 3 respectively. Using dimensionless values of the growth curve created for each homogenous unit and the average 24-hour rainfall in each rain gauge station, rainfall for different return periods were obtained. In addition to that, models were presented for calculating the mean rainfall. By comparing the values of rainfall in different return periods, the rainiest stations in short and long return periods were determined. In this research, the ratio between the average 24-hour rainfall and average annual rainfall was considered as a criterion in specifying stations with high flood risk. The stations with high flood risk may be given a higher priority in watershed management practices.