In the last two decades or so education and training have become central in the discussion of Britain's economic performance. There have been substantial changes in education and training policy (OECD 1995) and, at the same time, an expansion of participation in post-compulsory education. Both current and recent governments have aimed to raise work-based training through a mixture of persuasion and limited, selective, measures. Yet, there is considerable uncertainty about the extent to which the education and training system is enhancing skills in Britain or, for that matter, in many other countries. This paper concentrates on one dimension of the issue, namely work-based training statistics. There is uncertainty both over the amount of inputs into training and over its impact on skills. It argues that the uncertainty is, in part, the consequence of poor training statistics and that policy debate would benefit from improvements in the ways in which these statistics are compiled and interpreted.