Linkage analysis is generally carried out under single-gene models, while c
omplex traits are thought to be under the control of multiple interacting g
enes. Current issues related to linkage analysis for complex traits are dis
cussed. It is argued that linkage analyses should be carried out for sub-ph
enotypes, in addition to classical "affected-unaffected" phenotypes. Correl
ations for phenotypes among family members are often computed on the basis
of extreme phenotypes of a proband, which results in biased estimates. Meth
ods for ascertainment corrections are recommended. A generalized version of
heterogeneity analysis is introduced and are shown to provide an effective
single-locus analysis for complex traits.