Penetrance of schizophrenia and related disorders was calculated in 27 multiplex pedigrees ascertained by a consistent set of screening and selection criteria. The rationale for the study was that single major locus linkage models are frequently used on a pragmatic basis to analyze data for schizophrenia which is most likely to have a polygenic mechanism. Penetrance estimates assuming Mendelian inheritance represent maximum values and thus can provide guidance for construction of appropriate linkage models. Four diagnostic models were considered: narrow (schizophrenia and chronic schizoaffective disorder), intermediate (including other non-affective psychoses), broad (including schizotypal and paranoid personality disorders), and broad + suspected (including suspected schizophrenia spectrum disorders). Penetrance was calculated in the youngest affected adult sibship, under both dominant and recessive inheritance assumptions, either without correction, or with a correction that excluded individuals necessary to meet pedigree selection criteria. Without correction, penetrance values ranged from 0.70 to 0.90 assuming dominant and 1.0 to > 1.0 assuming recessive inheritance. After correction, the ranges were 0.30-0.51 for dominant and 0.47-0.59 for recessive models. The corrected values are likely to be overestimates given that the penetrance of any one locus in a multilocus model must be lower. It is suggested that lod score analyses of schizophrenia should attempt to derive information primarily from affected diagnoses, because information derived from unaffecteds under high penetrance models is likely to be spurious.