This paper provides an assessment of Goos and Manning’s (2007) polarised or ‘hourglass’ labour market thesis, which they claim has been caused by a period of routinisation where labour engaged in routine task occupations has been replaced by computer capital. It uses data taken from two waves of the National Child Development Study (NCDS) to study changes in labour market outcomes between 1981 and 2004 for a single cohort. While this dataset does demonstrate changes in employment consistent with routinisation, it is not clear that a polarising labour market is the inevitable conclusion. Looking at wage distributions for this cohort shows that the largest number of jobs continue to be in the middle of the wage spectrum. This paper questions the implicit assumption made by Goos and Manning (and subsequent authors) that initial wages provide a consistent proxy for job quality over the time period, and argues that the wage structure of occupations may have altered significantly over time.