Theoretical models predict that seed size and seed-bank persistence evolve interdependently, such that strong selection for one trait corresponds with weak selection for the other. This framework has been supported and rejected by empirical data, and thus, conclusive evidence is lacking. We expanded the seed size–persistence framework to include seed-coat thickness, a defence trait previously correlated with seed survival in soil. To do this, we used Abutilon theophrasti accessions with varied evolutionary histories and we quantified associations among seed traits including morphology, size, coat thickness, dormancy (percentage of viable seeds that fail to germinate under optimum conditions) and seed-bank persistence (percentage of viable seeds remaining after 1 year of burial). Statistical models were developed to test the hypothesis that combined measurements of seed-coat thickness and seed size better explain variability in seed-bank persistence than seed-size data alone. Results indicated that measurements of seed size (length, width, mass) were negatively correlated with coat:width ratio (coat thickness relative to seed width) and coat:mass ratio (coat thickness relative to seed mass). Accessions characterized by smaller seeds with proportionally thicker seed coats were more dormant and more persistent in soil than accessions characterized by larger seeds with proportionally thinner seed coats. Seed-coat thickness data improved the explanatory power of logistic regression models for seed-size effects on both seed-bank persistence and dormancy. These results indicate that supplementing seed-size data with seed-defence data may clarify previously reported contradictory results regarding trade-offs between seed size and seed-bank persistence.