There is general consensus that linkage with a quantitative trait locus cannot be detected with a reasonable number of sibpairs unless that trait has a very high heritability or the pairs are highly selected. However, the latter may require the screening of a very large number of pairs before a relatively small number of highly informative pairs is attained. At the same time, it is known that, for the same number of typed individuals, larger sibships tend to provide more linkage information than do independent sibpairs. We thus compared the efficiency of two sampling schemes, (1) random and (2) single ascertainment via an individual with an extreme phenotype, for sibship sizes ranging from two to five. The results demonstrate the clear trade-off between sibship size and the stringency of the ascertainment scheme.