Using data from the National Longitudinal Survey of Youth (NLSY) for the period spanning the years 1979-1991, this essay examines the impact of employer-provided formal training on the wage profile and on the mobility of young Americans making their transition to the labor market. By exploiting the longitudinal aspect of the data set, we are able to provide some control for unobserved individual and job-match heterogeneity by making use of the methodology proposed by Altonji and Shakotko (ReStud '87). The results show that (i) training with the current employer has a statistically and economically significant positive effect on the wage; (ii) employers seem to reward skills acquired through training with previous employers as much as skills they provide themselves; (iii) workers undergoing training have 18 percent lower starting salaries than other workers; this result is obtained by setting up a starting wage equation and by making use of a variable called on-the-job training still in progress at the time of the interview; (iv) with a hazard model which makes use of multiple employment spells by the same worker (thereby allowing the implementation of fixed-effects methods akin to the conditional logit method), skills acquired through formal training programs provided by the current employer seem to be fairly specific. The upshot from these results is that formal on-the-job-training in the current job contains both a general component which the employer rewards up to its market value and a specific component which reduces mobility while not being rewarded.