When editing a medieval document, the principles chosen for the transcription of the manuscript text must be deter- mined in accordance with the purpose of the transcription. One of the principal interests of this study is the question of how crusade sermon models were used and turned into actual sermons by preachers of the cross. It is, therefore, more relevant to reproduce a version of each model as it appears in a manuscript than to assemble an ‘authoritative’ text based on an ‘original’ version or the ‘best’ readings of all the different variants and strands of textual transmission. The idea is to obtain a version of each model such as would have been used by a medieval preacher. For most of the crusade sermon models collected here, the different manuscript versions show very few significant differ- ences that are not due to various types of scribal errors. Even where there exists a great number of manuscripts, which is the case for the models by James of Vitry, Gilbert of Tournai and Humbert of Romans, I have not been able to discern stemmatic traditions that, in terms of their content, yield significantly different versions of the models in question. The only exception concerns Bertrand de la Tour's models, which survived in three distinct versions of varying length.The texts reproduced here come close to what David'Avray has called a ‘critical transcription’.