Abstract: | This paper describes a study with two objectives. First, to test the effectiveness of a programme – carried out by teachers in natural contexts – for teaching the process of main idea identification. Second, to test whether the relationship between process and product variables are those expected according to the Van Dijk and Kintsch model (1983). In the first part of the study (Alonso-Tapia and Carriedo, in press) teachers had been trained in what to teach about main idea comprehension and how to teach it according to a model of teachers’ training. After receiving training, the experimental group of teachers (n = 11), who had improved more than the control group (n= 15) in knowledge of main ideas and of comprehension training strategies, had to apply the strategies and the instruction method learned, whilst teaching their students (6th, 7th and 8th grades). Pupils of experimental and of control teachers (n= 303 and 287 respectively) were assessed in order to test whether, after training, the former overtook the latter in main idea identification and in knowledge of the strategies related to this process. Results show that the experimental group of students performed better than the control group in the variables trained. Theoretical implications about the variables mediating the process of main idea identification are also discussed. |