Pure stands of a yellow rust-susceptible wheat cultivar, pure stands of a resistant cultivar, and a 1 : 1 random mixture of resistant and susceptible cultivars were compared to populations in which strips or hills of the cultivars were alternated to attain genotype units (units of the same host genotype) that were larger in area than that of a single wheat plant. These four host populations were grown in plots of different sizes in order also to alter the number of units per host population. The goal was to determine if increasing the number of genotype units in mixed populations of large genotype units improved disease control relative to pure-line populations by increasing the amount of inoculum exchange among genotype units. Random mixtures of the two cultivars always provided better disease control than did alternating strips or hills. Evidence for an effect of genotype unit number on the efficacy of mixtures for rust control was found in only one of three experiments. Random mixtures of the two cultivars increased grain yield relative to the pure stand mean, but alternating strips did not.