The recent rise in cohabitation in Britain is analysed using data from larg
e-scale surveys. There are major inconsistencies between different sources,
and retrospective estimates are higher than values reported at the time. R
etrospective data show markedly smaller numbers of cohabitation events just
before survey date. I discuss reasons for discrepancies and conclude that
no 'objective' measure of cohabitation exists and that comparison of differ
ent types of data requires care. I combine the data to produce a much large
r data set than hitherto available. Although cohabitation prevalence increa
sed substantially during the 1970s and 1980s, there was little change in su
ch characteristics as duration of cohabitation, ages of those cohabiting, a
nd whether it occurred before first marriage or ended in marriage or breakd
own. However, since the late 1980s, the average length of cohabitation has
increased markedly, which may indicate a qualitative change in the nature o
f cohabitation in Britain.