From 1975 to 2002, the proportion of people aged 15 to 59 who were neither students nor retired, and who were in employment, rose by 2.4% to 78.6%. This increase was uneven and depended both on short-term unemployment conditions and the continual rise in female economic activity. At the same time, throughout this period, employment within households became polarised: there were increasing numbers of households in which nobody or everybody worked, while the number of households in which only certain individuals worked fell. One explanation is linked to the changing make-up of households: in a quarter of a century, the proportion of households in which only one person may be able to work increased by 13%. However, whatever the household make-up, people have contributed to this polarisation. For example, among couples with one or more children, there are more households in which both parents work than in 1975, which is in line with the increasing number of women in the labour market. In order to analyse this phenomenon, Gregg and Wadsworth (1996, 2001, 2004) suggest the creation of a polarisation indicator which measures the disparity between a random distribution of employment within households and the real distribution. This article applies their method to the French data. We also see that the polarisation of employment in households has continued to increase since 1975, whatever the short-term employment conditions. Even when the rate of individual unemployment decreases (for example, in the second half of the 1980s or between 1997 and 2001), polarisation continues to increase. The first people to find work are therefore those who live in a household which already has experience of employment. Similar results can also be observed in most OECD countries: there has been a corresponding rise in households in which all or no adults work.