Since 2003 we have been involved in a project examining adult literacy and numeracy skills programmes based in the workplace and their impact on employees' skills levels and productivity. In this paper, we want to discuss two aspects of the dynamics of human capital in the context of work-based learning. The first is concerned with the impact learning can have on social capital within the firm and potential avenues by which social capital can affect productivity within the firm. The second is more concerned with how individuals' attitudes to learning both influence and are influenced by basic skills training. In terms of the first aspect of human capital, the ultimate purpose of our study is to answer the following questions: How might training affect a company's 'social capital'? Does the literature indicate how this might happen, and whether we might expect a significant impact on productivity? Can these effects be measured? In order to answer these questions, we attempted to discover whether the existing literature on 'social capital' could provide a framework for studying the relationship between 'social capital' and productivity at the level of the firm and for measuring either levels or changes in a robust fashion. We found that there was no such explicit framework, although there is a good deal of useful material. We have therefore used this literature to establish our own measurement framework, and have adapted and developed existing 'social capital' questionnaires for use at the level of the firm. This development work, together with some preliminary interview data, will be the main focus of the paper. Having said this, there is an important caveat. It is unlikely that a single model, or social capital questionnaire, "blueprint" can be perfect for use in all firms. As we have said, 'social capital' is, by its very nature, multifaceted and different aspects of trust, network and norms will be important in different firms and these aspects will interact in different ways. Therefore attempts to measure the quantity of 'social capital' in an organisation should be supported by qualitative investigations. Finally, it is clear that we need to be cautious when comparing firms that are different in terms of the social relations that aid organisation and production . Thus, the more specific knowledge we have about the firms we are studying, the more likely our measure of social capital is to reflect the reality of relations in the firm. With regard to the second aspect of human capital, we are also investigating the hypothesis that basic skills training at the workplace may affect learners' attitudes to learning or 'learning power', in addition to their (literacy or numeracy) skills, with further implications for the links between training and productivity. We are using a version of the Effective Lifelong Learning Inventory to measure employees' learning power, i.e. their attitudes towards learning and their motivation to learn. We believe that learning power may be important in this context in two ways: firstly, basic skills training may increase learners' learning power. We will be measuring this by examining any changes observed in the dimensions of learning power before and after a period of training. Secondly, learning power, if accurately measured by our instrument, should itself affect people's ability to acquire skills (both in terms of motivation and speed of acquisition). We should therefore expect to find a correlation between higher skills levels with higher scores on the ELLI dimensions. We therefore hypothesise that basic skills training may be particularly important as it may enhance learning power which itself will increases individuals' ability to acquire further skills. A feedback mechanism of this kind produces a dynamic that goes beyond the notion of a one-off improvement in individual skill levels.