Forecasts of the impact of computerisation on employment suggest that the likelihood that a given job will be replaced by computers is strongly linked to the proportion of routine tasks it involves. These forecasts often assume that machine-based jobs are at particular risk because they involve an especially high proportion of such tasks. In response to this assumption, the Labouring Capacity Index, based on the workforce survey jointly conducted by the German Federal Institute for Vocational Education and Training (BIBB) and the German Federal Institute for Occupational Safety and Health (BAuA), measures the proportion of non-routine tasks different types of work involve. It shows that workers in machine-based jobs also have to master demanding challenges in response to change, complexity and uncertainty, and hence these jobs involve a high proportion of non-routine tasks. In this article, we conduct a methodical assessment of the index by means of an ecological validation and explore some ways in which it could be refined or developed. We begin by presenting primary research on a car factory and, by way of contrast, a mould-making plant. Firstly, we compare the results of qualitative analyses of four production tasks with quantitative surveys conducted at the companies, using the Labouring Capacity Index. Secondly, we compare the results of these surveys with the data from the BIBB/BAuA workforce survey. This validation process confirms the validity of the Labouring Capacity Index, at least with respect to the production tasks analysed.