A method for decomposing partner availability into its demographic components (preferences, previous birth trends, migration and mortality, and structure by marital status) is presented and applied to marriage market estimates for selected census years 1911–91 in England and Wales. Preferences are a key component at the youngest ages. The role of other factors varies by age and time period. Contrary to widespread assumption, variation in cohort sizes resulting from past fertility trends is not the dominant contributor to partner availability during this period. Mortality and migration effects tend to be larger than the effect of birth trends and the two marital status components are generally the largest in size. Determinants of intercensal change are similar to the cross-sectional picture. Reasons for the modest contribution of trends in annual births are discussed. Cohort effects on partner supply are not necessarily absent but could arise through a number of mechanisms.