In 1992, DeVaull and King used the results of field tests to derive two similarity correlations for nonvolatile and volatile liquids, respectively, for the estimation of the rainout fraction x(R) of flashing liquid releases. These correlations are shown to be flawed, because an incorrect value of the liquid specific heat of CFC-11 has been used to derive them. The field test data corrected in 1994 with the UDM model are used to derive a new correlation valid for non-volatile as well as volatile liquids. The new correlation is x(R) = 0.6(1-3x(H)) where x(H) is the isenthalpic flash fraction. Use of the new correlation with the public domain consequence analysis code ALOHA is outlined. (C) 2008 Elsevier Ltd. All rights reserved.