We use British household panel data to explore the wage returns to training incidence and intensity (duration) for 6924 employees. We find these returns differ greatly depending on the nature of the training (general or specific); who funds the training (employee or employer); and the skill levels of the recipient (white or blue collar). Using decomposition analysis, we further conclude that training is positively associated with wage dispersion in Britain and a virtuous circle of wage gains but only for white-collar employees.