The on-going skilling debate has already yielded an abundance of contradictory theories, interpretations and empirical contradictions. Based on previous qualitative research in the Netherlands (Steijn and De Witte 1992), we have contributed to this debate by introducing the internal differentiation hypothesis. This paper addresses the empirical validity of this hypothesis by analysing data from a representative Dutch panel of 1,022 respondents in paid employment. The data show a small overall net upgrading trend. However, automation seems to have different effects for various occupational groups. For blue-collar workers, our findings suggest that a trend of internal differentiation does exist. Next, we examine the consequences (for underemployment and job satisfaction) of automation and changes in work content. Although our outcomes do not support our internal differentiation hypothesis, they do show the important effect of autonomy and complexity on feelings of underemployment. The outcomes justify Mottaz's statement that one of the consequences of this underemployment is job dissatisfaction.