This research investigated how computers might enable young learners to build models so that they can express and explore their ideas and hence they can gain understanding of the subject matter as well as developing modelling abilities. A design for a qualitative modelling environment was produced, which incorporated a simple rule-based metaphor that could be presented as a diagram. The design was founded on empirical evidence of children modelling as well as theoretical grounds. This research originated in and contributed to the Modus Project, a joint venture between King's College London and the Advisory Unit for Microtechnology in Education, Hertfordshire County Council. A prototype of the software, Expert Builder, was implemented by software engineers from the Modus team. The initial stage of evaluation, based on a questionnaire survey and widespread trialling, established that the tool could be used in a wide range of educational contexts. A detailed study of children using the qualitative modelling environment was conducted in three primary schools involving 34 pupils, aged nine to 11. They used the modelling environment within the classroom in their normal curriculum work over one school year on a variety of topics assisted by their class teacher. The modelling environment enabled cooperative groupwork and supported pupils in consolidating and extending their knowledge. A formative evaluation was used to inform the design of a revised version of the software. In addition the experiences of children using the software were analysed. A framework was developed which characterised the stages in the modelling process. Teachers in the study were observed to demonstrate the earlier stages of the modelling process and then to set tasks for the children based on the later stages of building and testing the models. The evidence suggested that the abilities to model were context dependent so that pupils as young as nine years old could undertake the whole modelling process provided that they were working on subject matter with which they were familiar. The teachers made use of computer based modelling in order to develop and reinforce pupils' understanding of various aspects of the curriculum and therefore they chose modelling tasks for the children. However in one school the children were given the opportunity to design and build models of their own choice and they demonstrated that they were able to carry out all the stages in the modelling process. A taxonomy of computer based modelling is proposed which could be used to inform decisions about the design of the modelling curriculum and could provide a basis for researchers investigating the modelling process. This would be useful for further research into the intellectual and social activities of people learning to model and for teachers seeking to develop a framework for the modelling curriculum. The National Curriculum (Department of Education and Science and the Welsh Office, 1990) specifies that early steps in computer based modelling should involve exploring models developed by others and pupils are not required to build models themselves until level 7 which is expected to be reached by more able 14 year-olds. In this thesis it is argued that a modelling curriculum should provide early opportunities for pupils to undertake the modelling process by developing simple models on familiar subject matter as well as opportunities for exploring more complex models as evidence from research reported in this thesis suggests that younger pupils are able to build models. In this way pupils will be enabled to acquire modelling capability as well as developing their understanding of a range of topics through modelling. Progression in modelling capability would involve constructing models of more complex situations and using a wider range of modelling environments.