We use important new training information from waves 8-10 of the British Household Panel Survey to document the various forms of work-related training received by men and women over the period 1998-2000, and to estimate their impact on wages. We initially present descriptive information about training: we find that most work-related training is viewed by its recipients as general, that the longest training courses are for induction purposes, that the vast majority of training takes place either at the workplace or at the employer’s training centre, and that most training is paid for by employers. We then estimate the impact of training – controlling for its financing method – on wages levels and wages growth. We find that employer-financed training increases wages both in the current and future firms, with some evidence that the impact in future firms is larger, especially for accredited training. These results are inconsistent with orthodox human capital theory with no credit constraints, but consistent with the relatively recent training literature on training in imperfectly competitive labour markets. They are also consistent with the hypothesis that firms offer credit-constrained workers binding training contracts whereby firms pay for general training and workers repay the ‘loan’ by receiving a post-training wage below their marginal product.