We study marital sorting on academic qualifications and latent ability in an equilibrium marriage market model using the 1972 UK Raising of the School-Leaving Age (RoSLA) legislation as a natural experiment that induced a sudden, large shift in the distribution of academic qualifications in affected cohorts, but plausibly had no impact on the distribution of ability. We show that a Choo and Siow (2006) model with sorting on cohort, qualifications, and latent ability is identified and estimable using the RoSLA-induced population shifts. We find that the RoSLA isolated low ability individuals in the marriage market, and affected marital outcomes of individuals whose qualification attainment were unaffected. We also decompose the difference in marriage probabilities between unqualified individuals and those with basic qualifications into causal effects stemming from ability and qualification differences. Differences in marriage probabilities are almost entirely driven by ability.