For the measurements of outcrossing rates in plant populations, current electrophoretic procedures permit many loci to be scored per individual progeny. Given that the total experimental effort or cost is limited, the choice exists then between assaying a large number of loci on a restricted number of individuals, or assaying a large number of individuals at a few loci. Using simple models and the criterion of minimising the variance of the estimate, several factors which affect this choice are considered (levels of polymorphism, heterozygosity, linkage disequilibrium, pollen or outcrossing heterogeneity). The general conclusion is that the actual level of outcrossing is a major factor in determining experimental strategy. Maximum efficiency for estimating outcrossing in predominantly inbreeding plants comes from large samples assayed for few polymorphic loci. In contrast, in predominantly outcrossing plants, more loci should be assayed at the expense of sample size for improved statistical efficiency.