Historique / Origines: Health promotion can assist schools to meet their targets in educational attainment and meet their social aims; young people that attend school have a better chance of good health and are likely to have better learning outcomes. Among the determinant of child health, the role of school was pinpointed especially teaching practices. Objectifs: This communication aims to provide knowledge on the process which conducts a teacher to take into account health promotion issues into his teaching practice. Valeurs et principes sous-jacents: Health promotion values and principles Fondement de connaissance/Fondement de preuve: Our framework is based on health promotion effectiveness and health promoting school (Barnekow & al., 2006; St Leger & al., 2007). Contexte d'intervention/projet/travail: This study is based on the analysis of a health promoting programme based on teacher training and school staff support, which was implemented in 22 French primary schools (2003-2007), involving 116 teachers. Méthodes: The investigation tool is a questionnaire entitled: practice and representation in the field of health education (HE). A first study based on the data collected during the third year of the programme was done (Simar & Jourdan, 2009). It showed that three factors were significantly linked to health promoting practices: training, collaborative work and high level of individual interest in the field of HE. The present work aims at identifying the way in which teaching practices have changed along the three years of the implementation. Generalized Estimating Equations (GEE) have been used in order to analyze repeated measures of discrete data. The dependant variable in this study is health promoting teacher practices(HPP). Résultats et Conclusions: All thing being equal, the GEE show that six variables are linked with HPP over the years. Among them, four factors are categorical variable: collaborative work (OR=3,13/p-value=1.10-4), single grade class (OR=2,50/p=1,6.10-3), small school (OR=5,66/p=1,1.10-3) and length of teachers involvement in the programme (OR=4,29/p=3.10-3). In addition, two factors are continuous variable; there are training (OR=1,07/p<10-4) and individual interests in the field of HE (OR=1,41/p=5.10-3). The odds ratios were measured for one unit of training or one unit of individual interest. For example, it means if a teacher had completed the training (21h) during the first year, he was 3.82 times as likely to have HPP. This study gives us knowledge on the content of the training which should be delivered and some information on the way to manage a health promoting programme.