Flowerdew and Dudley-Evans (2002) described a prototypical structure for decision letters based on a personal database of letters written by one editor for the journal "English for Specific Purposes." In this article, I analyse a publicly available corpus of 59 decision letters from 48 different editors of a wide range of scientific journals and find that the Flowerdew and Dudley-Evans prototypical structure requires modification to be applicable to this broader spectrum of decision letters. In addition, all but two of the linguistic features of decision letters described by Flowerdew and Dudley-Evans (2002) are used by less than 15 per cent of the editors in the present corpus. Three previously undescribed linguistic features, that are common to more than 70 per cent of the editors, are identified in this study.