It is unresolved what, if any, characteristics should be used as a basis for assigning psychotic probands to different liability classes in high density family studies seeking to detect possible genetic linkage. Justification for any such assignments should ideally ensue from empirical evaluation of unselected samples. It has been suggested that the genetic liability of probands with an "organic" psychosis is lower than that found in "primary" cases. Should such cases be assigned differential liabilities in linkage analyses as one way of modeling etiologic heterogeneity? Utilizing data from a population-based family study conducted in County Roscommon in Western Ireland, we examined risk in the relatives of psychotic probands as a function of clinician ratings reflecting the probability that the proband's illness was organic. Contrary to expectation, risk was not significantly lower in relatives of probands whose illness was rated as organic by experienced clinicians. Attempts to identify possible phenocopies of psychosis with a lower familial liability in this treated epidemiologic sample were unsuccessful.