We aimed to model binding of donor antibodies to virus that infects COVID-19 patients following transfusion of convalescent plasma (CCP). An immunosorbent assay was developed to determine apparent affinity (Kd, app). Antibody binding to virus was modelled using antibody concentration and estimations of viral load. Assay and model were validated using reference antibodies and clinical data of monoclonal antibody therapy. A single Kd, app or two resolvable Kd, app were found for IgG in 11% or 89% of CCP donations, respectively. For IgA this was 50%-50%. Median IgG Kd, app was 0.8nM and 3.6nM for IgA, ranging from 0.1-14.7nM and 0.2-156.0nM respectively. The median concentration of IgG was 44.0nM (range 8.4-269.0nM) and significantly higher than IgA at 2.0nM (range 0.4-11.4nM). The model suggested that a double CCP transfusion (i.e. 500 mL) allows for > 80% binding of antibody to virus provided Kd, app was < 1nM and concentration > 150nM. In our cohort from the pre-vaccination era, 4% of donations fulfilled these criteria. Low and mid-range viral loads are found early post exposure, suggesting that convalescent plasma will be most effective then. This study provides a biochemical rationale for selecting high affinity and high antibody concentration CCP transfused early in the disease course.