It has frequently been suggested in the literature that a polycentric distribution of employment and people shortens commuting distances because people locate within or close to their employment sub-centre (the co-location hypothesis). Having studied the three biggest French metropolitan areas over the past decade it has been established that co-location affects only a minority of inhabitants, of whom there are fewer in 1999 than there were nine years earlier. Indeed, the majority of people living in a sub-centre work outside their sub-centre of residence. This situation was even more marked in 1999 than it was in 1990. In addition to this, the majority of jobs located in sub-centres are held by non-residents who are generally living further and further from their place of work.