CAD has been highlighted as a core technology in the development of flexible automation. However, experience during the 1980s has led to a great deal of scepticism over the benefits to be derived from adopting the technology. This article uses case study evidence to suggest that many of the problems result from a failure to adequately address the work organisation issues highlighted by CAD. The broad conclusion is that the full innovative potential of CAD has yet to be realised. In part this is because management responses to issues of organising and controlling work have been limited by traditional assumptions and beliefs, in turn often sustained by existing organisational arrangements and measures of effective performance. To realise the full potential of CAD as a core element of flexible automation a more ‘human-centred’ approach may be required with regard to questions of job design and supervision.