The difficulties in learning and teaching astronomy, including the use of models, have been well documented. Models, when used by teachers, are usually utilised in a transmissive way to impart content to pupils. The process of constructing mental models, an essential aspect of the development of astronomy knowledge (both historically and today) is usually overlooked. This research explored how models can promote mental model building in classrooms. Mental models are defined as human cognitive constructions which are used to describe and explain phenomena which cannot be experienced directly. They are used both to understand the world and to articulate understandings to others. Models may be scaled miniatures or scaled enlargements or working models, each of which represents a mental model in some, but not all, its properties. An example is an orrery, which is a representation of part of the solar system. This study investigated how pupils can be helped to construct the scientists’ mental model of the Sun-Earth-Moon system. It used an approach suggested by Hesse (1966) in which a model is repeatedly critiqued with respect to the extent that it both represents and fails to represent the structure and properties of a mental model. The study proposed that the process of repeated critiques would assist learners to construct and consolidate the scientists’ mental model of the Sun-Earth-Moon system. A research programme was designed to investigate whether pupils could be assisted to construct the scientists’ mental model of the Sun-Earth-Moon system within the constraints of an ordinary classroom by using the process of critiquing a model of the scientists’ mental model. A 12-lesson teaching intervention, conducted over a period of three weeks, was designed for that purpose. The teaching strategy, which was developed from well-trialled constructivist research, consisted of four phases: Focus on the Mental Model; Mental Model Building and Critiquing; Using the Mental Model to Solve Problems; and Reflection. The study reported interviews with twelve intermediate school teachers to ascertain their views about the challenges of teaching astronomy. This was followed by an initial classroom pilot intervention and then a formal classroom trial intervention with Year 7-8 pupils in the same intermediate school in New Zealand. Both classes were similar and both were taught by teachers in their second year of teaching. An orrery was used as the model of the mental model of the Sun-Earth-Moon system in both the pilot and trial interventions. Insights learned from the pilot intervention were used to modify the format of the trial intervention, which took place one year later. Both the pilot intervention and the trial intervention were monitored by: pre- and post-interviews with Year 7-8 pupils; interviews with the intervention teacher after most of the lessons; pre-, post- and post-post written surveys of all members of the class; and audio-tapes of the lessons. The pre-interviews and pre-surveys showed that the pupils began both interventions with a range of mental models of the Sun-Earth-Moon system. Furthermore, their other concepts (about tides, phases of the Moon, spatial relationships, etc.) were often apparently unrelated to the Sun-Earth-Moon system or to each other. The findings showed that most of these Year 7-8 pupils were enabled to build the scientists’ mental model of the Sun-Earth-Moon system and that the use of models is effective in promoting mental model building. As well as acquiring these understandings about processes in astronomy, the pupils also achieved the traditional goals in astronomy knowledge and concepts (although concepts associated with the Moon still proved difficult for these pupils, as did an understanding of Earth’s axis of rotation). The pupils responded positively to the approach, they did not tire of the repeated critiquing of the model and they developed their ideas about models in science at large. The study demonstrated that all four phases of the modified teaching strategy were necessary for the mental model to be both constructed and consolidated. It is also clear that pupil interactions, both in small groups and in the whole class, were important factors in the mental model building process. In addition, the study further indicated that, contrary to some other research, Year 7-8 may be an appropriate age to begin teaching about models in science in general.