Broadly there are three alternative sampling strategies for linkage studies of complex disorders such as schizophrenia. These are to select either affected sib pairs (and their parents) or small-to-medium-size pedigrees, or large pedigrees. The advantages and disadvantages of each are discussed. Our studies show that, even in the presence of heterogeneity, linkage can be found in clinically realistic sample sizes of nuclear families or medium-sized pedigrees. However, relying on affected sib pair methods in less satisfactory.