The runoff coefficient φ is a crucial parameter for the estimation of the mean value of annual maximum flood peak discharges in ungauged watersheds, where no direct measures are available. If the rational method is applied as a rainfall–runoff transformation model, the runoff coefficient accounts for all the hydrological losses, and it can be conceptually defined as the fraction of the total rainfall contributing to the flood peak response. In the present paper, focusing on Southern Peninsular Italy, a regression model is proposed to improve the prediction of the above defined runoff coefficient as a function of several parameters describing both morphological and mean annual climatic watershed characteristics. Morphological features are described by using the Soil Conservation Service permeability classification and the related variable S, referred to an average antecedent moisture condition. Different climatic indices enable a subsequent enhancement of S in order to account for the specific mean annual moisture condition of each watershed.