Flood is one of the events that has attracted attention of hydrologists. In this work, one of the important indices of flood, the maximum-daily mean-discharge, is determined for some western Iran watersheds of indices: Gamasiab, Qaresoo, Seymareh, Kashkan, Sezar and Abshineh.

Daily mean data was chosen from stream-gauging stations and a 30-year common period was selected and flood frequency analysis was done using HYFA and TR software and the best distribution was chosen using the goodness of fit tests. Then, discharge values with different return periods of 2, 5, 10, 25, 50, 100, 500 and 1000 years were evaluated. Modeling was done using the regional analysis by multiple regression technique between maximum-daily mean-discharge and physiographic characteristics of the basins. The main parameter for choosing the model was the adjusted coefficient of determination, and the meaningful level was standard error, and observed discharger was proved by computed discharge plot. So, different models with different parameters were selected from the power, exponential, linear and logarithmic models. But the power model was the one that was used more than the others. According to this work, the main parameters that affect discharge are: channel length, drainage density and time of concentration. Considering the errors, it is concluded that increasing the return period results in an increase in the model error, in a way that for a 1000-year return period, the error reaches 32.2%.