This paper applies the familiar theoretical distinction between general and specific training to the empirical task of estimating the returns to in-company training. Using a firm-level dataset which distinguishes between general and specific training, we test for the relative effects of the two types of training on productivity growth. We find that although general training has a statistically positive effect on productivity growth, no such effect is observable for specific training. This positive effect of general training remains when we control for factors such as changes in work organisation and corporate re-structuring, firm size and the initial level of human capital in the enterprise. Moreover, the impact of general training varies positively with the level of capital investment.