The need for a continued reassessment of reservoir safety in relation to heavy rainfall and floods is of paramount importance. The latest advice given by the Institution of Civil Engineers states that for much of Somerset in southwest England, the Flood Studies Report underestimates the frequency of heavy rainfall and that local data should be adopted. However, this advice has not always been followed. Furthermore, the advice does not extend to other areas in southwest England. Therefore, it is essential to assess in detail the reliability of the four methods which can be used in order to estimate storm rarity at present in southwest England. These are the Bilham method, the Flood Studies Report (FSR) method, the FORGE method, and the Four Parameter Model. Both the Flood Studies Report and the FORGE method give higher values of storm rarity than either the Bilham or Four Parameter Model. The Bilham method is not sensitive to geographical variation in rainfall frequency. Comparison of all four methods with observed storm events and storm depths showed that the Four Parameter Model gave the most realistic results, followed by the Bilham method. For a given storm rarity the Four Parameter Model tends to give the highest storm depth, with the differences increasing with storm rarity. These new results suggest that in southwest England many existing dams and other drainage works have been underdesigned.