A methodology was developed to investigate the question: 'What facilitates pupil learning in classrooms?' The question was interpreted to constrain the methodology in requiring: a naturalistic orientation, a focus on the range of variables influencing pupil experience of content, and a non-selective observation procedure. These constraints led to a study of the experience of three case study pupils during their in-class opportunity to interact with the content of an integrated unit. A multi-method approach was adopted in order to obtain as much information as possible about pupil experience of content. Three observers alternated in continuously observing and recording the behaviours of the three pupils. Audio-recordings were made of public lessons and audible pupil interactions were noted. Copies of all teacher resources and pupil work were taken. Teacher and observer perceptions were noted. The pupils were tested before the seven week unit, immediately afterwards, and again one year later. After the long-term post test, interviews were conducted to obtain: information about the validity of the tests, more information about pupil experience of tested content, and pupil perceptions of their learning. The test was sub-divided into three parallel sets of items. For each of these items, files were constructed for the case study pupils which included descriptions of their total in-class opportunity to interact with this content. The teacher and researcher engaged in a prediction procedure using pretest responses and records of pupil opportunity to interact with content. This was done to identify extant insight into pupil learning and to reveal researcher bias which might unduly restrict the data analyses. This procedure showed that both teacher and researcher were able to successfully predict over 70 % of pupil outcomes from the observational record. Factors which inhibited higher prediction success rates for both teacher and researcher were failure to: recognize the primacy of concrete experience as a pre-requisite for long-term learning, identify the presence, persistence, and retro-active effects of pupil misconceptions, account for the strength of the relationship between time and learning, and take into account out-of-class learning opportunities on in-class achievement. The item files were classified according to the five possible outcomes: already known, learned and remembered, learned and forgotten, not learned, and mislearned. The average time spent by pupils on content which was not learned was less than one fifth of the time spent on content learned and remembered. The number of occasions upon which the opportunity to interact with content occurred was also shown to be related to pupil learning with content learned and remembered occurring over more than six school days. An analysis of pupil behaviour was carried out in relation to the five learning outcomes. Mean frequencies and rates per hour were calculated in order to show which behaviours were related to learning because they took up the time spent, which behaviours were related to learning over and above the time spent, which behaviours inhibited learning, and which behaviours were unrelated to pupil learning. The consistency of these relationships across the three data sets was also calculated. Opportunity to interact with content both in teacher-directed lessons and in self-directed or peer-directed task contexts was found to be related to learning. Opportunity to attend to a concrete demonstration of new concepts, and opportunity to attend to teacher-pupil discussion were found to be systematically related to pupil long-term learning. Picture attending opportunity and involvement in mime were found to be related to short-term learning. Rubbing out during individual tasks was found to be consistently related to pupil learning. Peer interaction was found to relate, to pupil learning both positively and negatively depending upon task contexts and the rate and type of communication. There were differences between the case study pupils in their overt involvement with content, their perceptions of task requirements, and their ability to obtain useful information from T. An analysis of exceptions to the overall pattern of findings revealed that chart attending opportunity and diagram drawing behaviour occurred at very high rates for content which was learned in spite of less time spent. Pupil concrete experience and pupil misconceptions were found to be confounding variables in relation to the time spent. The case study pupils were found to have responded in a characteristic manner to new content with high rates of fiddling, talking to self, and peer interaction. These behaviours are postulated to indicate the occurrence of cognitive conflict and cognitive restructuring in covert pupil processing. The study findings were used to generate a grounded nascent theory of the learning of the case study pupils. In this theory pupil learning is viewed as the effect of interaction between three major variable clusters: facilitative opportunity to learn, pupil behaviour, and pupil resource access. Given the interaction of the three variable clusters the nature of pupil learning is explained as a three stage process which occurs over time. The first and briefest stage (awareness and disequilibrium) occurs when a pupil actively engages with new content. The more relevant prior knowledge and experience the pupil has, the more likely the engagement in this initial stage of the learning process. The second stage (cognitive restructuring, resolution, and integration) occurs as the child restructures existing schema, and attempts to forge links between existing knowledge and new content. The final stage (schema development and anchoring) involves the accretion of enough appropriate links to make the new information part of the pupil's general knowledge of the world. It is claimed that the findings of this study have important implications for research but they are neither generalizable to different contexts, nor prescriptive for practitioners.